Systems and methods for active light-based precise localization of aircraft in GPS-unavailable environments

Abstract

A system and method for providing guidance to assist an eVTOL airborne vehicle in performing landing and takeoff operations at a landing site in a GPS-unavailable environment is disclosed. An exemplary system includes an airborne vehicle including a camera configured to generate images based on information transmitted from a plurality of light sources disposed adjacent a landing surface of the airborne vehicle, and a controller circuit configured to receive the generated images and determine a position and orientation of the airborne vehicle based on the received images, wherein the light sources are arranged in a predetermined pattern on the landing surface, and a characteristic of the light emitted from each light source is modulated with respect to time.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This disclosure claims priority to U.S. patent application Ser. No. 18 / 451,055 (Attorney Docket No. 16163.0047-00000), filed on August 16, 2023, entitled "System and Method for Active Light-Based Precision Localization of Aircraft in GPS-Non-Available Environments," which claims priority to U.S. patent application Ser. No. 18 / 451,055 (Attorney Docket No. 16163.0047-00000), filed on October 30, 2022, entitled "System and Method for Active Light-Based Precision Localization of Aircraft in GPS-Non-Available Environments." This application claims priority to U.S. Provisional Patent Application No. 63 / 420,616 (Attorney Docket No. 16163.6001-00000) and U.S. Provisional Patent Application No. 63 / 381,571 (Attorney Docket No. 16163.6001-01000), entitled "System and Method for Active Light-Based Precision Localization of Aircraft in GPS-Unavailable Environments," filed October 31, 2022, the contents of which are incorporated herein in their entirety for all purposes.

[0002] The invention described in this patent application and its various embodiments were made at least in part through Department of Defense support (Contract No. FA8649-22-P-0797). To correct a clerical error, the words "FA8649-21-P-0038" in two previously filed provisional patent applications (Serial Nos. 63 / 420,616 and 63 / 381,571) are to be substituted and are hereby replaced with "FA8649-22-P-0797." The United States Federal Government may retain certain license rights in this invention.

[0003] Technical Field This disclosure relates generally to the field of powered aerial vehicles. In particular, without limitation, this disclosure relates to electric vertical take-off and landing (eVTOL) aerial vehicles and methods for providing high-precision, high-reliability active light-based landing and take-off positioning guidance therefor. Certain aspects of this disclosure relate generally to precision landing and take-off systems that can be used with other types of vehicles, but provide particular advantages in aerial vehicles. Summary of the Invention

[0004] Embodiments of the present disclosure generally relate to the field of electric vertical take-off and landing (eVTOL) airborne vehicles. Further, without limitation, the present disclosure relates to systems and methods for providing guidance to assist in take-off and landing operations of eVTOL airborne vehicles in environments where GPS is unavailable or in areas where GPS is degraded and accuracy is limited. The present disclosure also relates to methods for providing take-off and landing guidance and estimating the pose of the airborne vehicle relative to a landing surface. These methods may include utilizing an active constellation of fiducial light sources in the infrared or visible spectrum distributed at known, fixed locations around a designated landing site. These light sources are viewed by an onboard camera as the vehicle approaches the landing site. Patterns from the light sources projected onto the camera's image plane can be used to reliably calculate the camera's pose (position and attitude) with the appropriate level of accuracy required for precision eVTOL landing.

[0005] One aspect of the present disclosure relates to a system that includes a landing surface of an airborne vehicle, the landing surface may include a plurality of light sources arranged in a predetermined pattern, wherein a characteristic of the light emitted from each light source is configured to be modulated over time.

[0006] Another aspect of the present disclosure relates to an airborne vehicle that includes a camera configured to generate images based on information transmitted from a plurality of light sources positioned near a landing surface of the airborne vehicle, and a controller circuit configured to receive the generated images and determine a position and orientation of the airborne vehicle based on the received images, wherein the light sources are positioned in a predetermined pattern on the landing surface, and a characteristic of the light emitted from each of the light sources is modulated with respect to time.

[0007] Another aspect of the present disclosure relates to a system including a plurality of light sources disposed on a landing surface of an airborne vehicle, the arrangement of the light sources defining a set of intersecting imaginary lines, the light sources being disposed on each imaginary line, and the distances between adjacent light sources on each imaginary line being unequal.

[0008] Another aspect of the present disclosure relates to a method for estimating the pose of an airborne vehicle. The method may include providing a landing surface including light sources arranged in a predetermined pattern, modulating characteristics of light emitted from the light sources over time, receiving input signals related to the light emitted from the light sources using a camera mounted on the airborne vehicle, generating images of the light sources based on the received input signals, and determining a position and orientation of the airborne vehicle based on the images. Determining the position and orientation of the airborne vehicle includes detecting at least one of the light sources in the image, determining which of the at least one light sources arranged in the predetermined pattern the detected light source is, and determining the position and orientation of the airborne vehicle based on determining which of the at least one light sources arranged in the predetermined pattern the detected light source is.

[0009] Another aspect of the present disclosure relates to a computer-implemented system for estimating the pose of an airborne vehicle. The system may include a landing surface including light sources arranged in a predetermined pattern and at least one processor. The processor may be configured to modulate characteristics of light emitted from the light sources with respect to time, activate a camera mounted on the airborne vehicle to receive input signals related to the light emitted from the light sources, cause the camera to generate images of the light sources based on the received input signals, and determine a position and orientation of the airborne vehicle based on the generated images. Determining the position and orientation includes detecting at least one light source in the image, determining which of the at least one light sources arranged in the predetermined pattern the detected light source is, and determining the position and orientation of the airborne vehicle based on the determination of which of the at least one light source arranged in the predetermined pattern the detected light source is.

[0010] Another aspect of the present disclosure relates to a computer-implemented method for estimating a pose of an airborne vehicle, the method, when executed by at least one processor, including acts of modulating, with respect to time, characteristics of light emitted from light sources arranged in a predetermined pattern on a landing surface of the airborne vehicle, activating a camera mounted on the airborne vehicle to receive an input signal related to the light emitted from the light sources, enabling the camera to generate an image of the light sources based on the received input signal, and determining a position and orientation of the airborne vehicle based on the image. Determining the position and orientation includes detecting at least one of the light sources in the image, determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is, and determining the position and orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is.

[0011] Another aspect of the present disclosure relates to a non-transitory computer-readable medium storing a set of instructions executable by at least one processor of a device to cause the device to perform a method, the method including: modulating, with respect to time, characteristics of light emitted from light sources arranged in a predetermined pattern on a landing surface of the airborne vehicle; activating a camera mounted on the airborne vehicle to receive input signals related to the light emitted from the light sources; enabling the camera to generate an image of the light sources based on the received input signals; and determining a position and orientation of the airborne vehicle based on the image. Determining the position and orientation includes detecting at least one of the light sources in the image, determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is, and determining the position and orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is.

[0012] Another aspect of the present disclosure relates to an airborne vehicle. The airborne vehicle may include a camera configured to generate an image based on information received from a plurality of light sources disposed on a landing surface of the airborne vehicle, and a processor associated with the camera. The processor may be configured to receive the image and perform the following operations: detect light sources in the image using a detection algorithm, the light sources being disposed on the landing surface and configured to emit light detectable by the camera; associate locations in the image representing the detected light sources with corresponding locations of the light sources on the landing surface, the processor being configured to perform the association in a first mode of operation and a second mode of operation; execute one or more association algorithms in the first mode of operation to generate an association confidence score; execute one or more tracking algorithms in the second mode of operation based on the confidence score obtained from the first mode of operation; and determine a position or orientation of the airborne vehicle based on the performed association.

[0013] Another aspect of the present disclosure relates to a method for operating an airborne vehicle, the method including generating an image with a camera based on information received from a plurality of light sources located on a landing surface of the airborne vehicle, detecting the light sources in the image using a detection algorithm, the light sources being disposed on the landing surface and configured to emit light detectable by the camera, associating locations in the image representing the detected light sources with corresponding locations of the light sources on the landing surface, the associating step including a first mode of operation and a second mode of operation, the first mode of operation including executing one or more association algorithms and generating an association confidence score, and the second mode of operation including executing one or more tracking algorithms based on the confidence score obtained from the first mode of operation, and determining a position or orientation of the airborne vehicle based on the performed association.

[0014] Another aspect of the present disclosure relates to a navigation system for an airborne vehicle, the navigation system including: a camera configured to generate images based on information received from a plurality of light sources arranged in a predetermined pattern on a landing surface of the airborne vehicle; and a processor associated with the camera and configured to receive the images and perform the following operations: using the processor to activate the camera mounted on the airborne vehicle to receive input signals related to light emitted from the light sources arranged in the predetermined pattern on the landing surface of the airborne vehicle, the light having a time-modulated characteristic; enabling the camera to generate at least two images of the light sources based on the received input signals; detecting the light sources in the at least two images using a detection algorithm; and associating locations in the images representing the detected light sources with corresponding locations of the light sources on the landing surface, the processor being configured to perform the association in a first mode of operation and a second mode of operation; executing one or more association algorithms in the first mode of operation; executing one or more tracking algorithms in the second mode of operation based on results obtained from the first mode of operation; and determining a position or orientation of the airborne vehicle based on the performed association.

[0015] Another aspect of the present disclosure relates to a system that may include a landing surface for an airborne vehicle and a plurality of light sources arranged in a predetermined pattern, wherein a characteristic of light emitted from each light source is configured to be modulated with respect to time, the plurality of light sources including a linear light source and a point light source, and the landing surface including a portable landing surface. [Brief explanation of the drawings]

[0016] [Figure 1A] 1 shows a schematic of a conventional instrument landing system (ILS) that provides horizontal and vertical guidance to guide an aircraft along a runway.

[0017] [Figure 1B]Shown is the emission of localizer and glideslope signals that assist the pilot with horizontal and vertical guidance for landing. [Figure 1C] Shown is the emission of localizer and glideslope signals that assist the pilot with horizontal and vertical guidance for landing.

[0018] [Figure 2] 1 illustrates a schematic diagram of an exemplary landing / takeoff approach using optical navigation in a GPS-unavailable environment, in accordance with a disclosed embodiment.

[0019] [Figure 3] 1 illustrates an exemplary precision landing and takeoff system and data communication paths between an aircraft, a vertiport, and a control unit according to disclosed embodiments.

[0020] [Figure 4A] 1 illustrates an exemplary landing surface or vertiport according to some disclosed embodiments. [Figure 4B] 1 illustrates an exemplary light source according to some disclosed embodiments.

[0021] [Figure 5] 1 illustrates a focal plane array (FPA) camera image of an exemplary vertiport light source, according to some disclosed embodiments.

[0022] [Figure 6] 1 illustrates an exemplary overview of the algorithms and data pipelines in operation of a precision landing and takeoff system according to some disclosed embodiments.

[0023] [Figure 7] 1 illustrates an exemplary detection algorithm for locating a light source in an image captured by a camera, according to some disclosed embodiments.

[0024] [Figure 8]10 shows a data plot illustrating a comparison of a signal before and after signal processing using a bandpass filter, according to some disclosed embodiments.

[0025] [Figure 9] 10 shows a data plot illustrating a comparison of a signal before and after a Discrete Fourier Transform (DFT) calculation, according to some disclosed embodiments.

[0026] [Figure 10A] 10 shows simulation data plots illustrating the effect of the number of light sources on horizontal localization accuracy, according to some disclosed embodiments. [Figure 10B] 10 shows simulation data plots illustrating the effect of the number of light sources on vertical localization accuracy, according to some disclosed embodiments.

[0027] [Figure 11A] 10 shows simulation data plots illustrating the effect of light source constellation size on horizontal localization accuracy, in accordance with some disclosed embodiments. [Figure 11B] 10 shows simulation data plots illustrating the effect of light source constellation size on vertical localization accuracy, in accordance with some disclosed embodiments.

[0028] [Figure 12A] 10 shows simulation data plots illustrating the effect of centroid error on algorithm robustness in the horizontal direction, according to some disclosed embodiments. [Figure 12B] 10 shows simulation data plots illustrating the effect of center of gravity error on algorithm robustness in the vertical direction, according to some disclosed embodiments.

[0029] [Figure 12C]10 illustrates a simulated data plot showing horizontal localization error relative to a landing trajectory during a simulated approach of an aircraft, according to some disclosed embodiments. [Figure 12D] 10 illustrates a simulated data plot showing vertical localization error relative to a landing trajectory during a simulated approach of an aircraft, according to some disclosed embodiments.

[0030] [Figure 13A] 1 illustrates an exemplary matching and association process using a thin-plate spline algorithm, according to some disclosed embodiments. [Figure 13B] 1 illustrates an exemplary matching and association process using a thin-plate spline algorithm, according to some disclosed embodiments. [Figure 13C] 1 illustrates an exemplary matching and association process using a thin-plate spline algorithm, according to some disclosed embodiments. [Figure 13D] 1 illustrates an exemplary matching and association process using a thin-plate spline algorithm, according to some disclosed embodiments.

[0031] [Figure 14A] 1 illustrates an exemplary camera image with detected points, in accordance with some disclosed embodiments.

[0032] [Figure 14B] 1 shows a schematic diagram of normalized detection points in a predefined space, in accordance with some disclosed embodiments;

[0033] [Figure 14C] 4C illustrates an exemplary Hough transform space including a mapping of a line passing through the particular point of FIG. 4B, according to some disclosed embodiments.

[0034] [Figure 14D]14D shows a schematic diagram of a line formed by discretizing the points of FIG. 14C in Hough space, according to some disclosed embodiments.

[0035] [Figure 14E] 1 illustrates an exemplary Hough transform space with line refinement, according to some disclosed embodiments.

[0036] [Figure 14F] 14B illustrates an exemplary k-means clustering representation of the refined lines of FIG. 14E, according to some disclosed embodiments.

[0037] [Figure 14G] 1 illustrates an exemplary representation of mapped points in two-dimensional space, according to some disclosed embodiments.

[0038] [Figure 14H] 14G illustrates an example projection of the points of FIG. 14G onto integer grid space, according to some disclosed embodiments.

[0039] [Figure 14J] 14H illustrates an exemplary detection with associated annotation of light sources based on the information in FIG. 14H, according to some disclosed embodiments.

[0040] [Figure 15A] 10 shows an example image illustrating shifts between frame points and corresponding new predicted positions using homography, in accordance with some disclosed embodiments.

[0041] [Figure 15B] 10 illustrates an example image generated using a nearest neighbor-based search to identify detections corresponding to tracked known associations, in accordance with some disclosed embodiments.

[0042] [Figure 16A] 1 illustrates a linear pattern of an exemplary constellation of light sources, according to some disclosed embodiments.

[0043] [Figure 16B] 1 illustrates an exemplary constellation star pattern of a light source, according to some disclosed embodiments.

[0044] [Figure 17A] 1 illustrates an exemplary constellation pattern of a light source, according to some disclosed embodiments. [Figure 17B] 1 illustrates an exemplary constellation pattern of a light source, according to some disclosed embodiments.

[0045] [Figure 18A] 1 illustrates an example set of lines using a Random Sampling Consensus (RANSAC) sampling method, according to some disclosed embodiments. [Figure 18B] 1 illustrates an example set of lines using a Random Sampling Consensus (RANSAC) sampling method, according to some disclosed embodiments.

[0046] [Figure 18C] 10 illustrates a visualization of the process of computing inliers after finding a set of suitable candidate lines, according to some disclosed embodiments.

[0047] [Figure 18D] 10 illustrates an example of using angle crossing ratios to determine the identity of lines in a constellation, according to some disclosed embodiments.

[0048] [Figure 19] 1 illustrates an exemplary voting scheme using linear crossover ratios to determine point identity, according to some disclosed embodiments.

[0049] [Figure 20A]1 is a flowchart illustrating an exemplary method for pose estimation using a data association algorithm, in accordance with some disclosed embodiments.

[0050] [Figure 20B] 10 shows an example plot illustrating a simulated circular orbit flying around a constellation, in accordance with some disclosed embodiments.

[0051] [Figure 21] 21A and 2100B show data plots 2100A and 2100B illustrating altitude estimates and corresponding errors for the flight trajectory shown in FIG. 20, according to some disclosed embodiments.

[0052] [Figure 22] 22A and 22B show data plots 2200A and 2200B illustrating north estimate versus ground truth and corresponding error for the flight trajectory shown in FIG. 20, according to some disclosed embodiments.

[0053] [Figure 23] 23A and 23B show data plots 2300A and 2300B illustrating east estimate versus ground truth and corresponding error for the flight trajectory shown in FIG. 20, according to some disclosed embodiments.

[0054] [Figure 24A] 1 shows a schematic of an exemplary random dot marker according to some disclosed embodiments.

[0055] [Figure 24B] 1 illustrates an exemplary point identification approach using a Locally Likely Clearance Hashing (LLAH) algorithm, according to some disclosed embodiments. [Figure 24C] 1 illustrates an exemplary point identification approach using a Locally Likely Clearance Hashing (LLAH) algorithm, according to some disclosed embodiments.

[0056] [Figure 24D] 1 illustrates an exemplary crossover ratio discretization method for creating a discretized crossover ratio sequence, in accordance with some disclosed embodiments.

[0057] [Figure 24E] 1 illustrates an exemplary hash table for keypoint registration, according to some disclosed embodiments.

[0058] [Figure 25] 10 illustrates an exemplary area intersection ratio calculation for a subset of coplanar lights, according to some disclosed embodiments.

[0059] [Figure 26A] 10 illustrates exemplary waveforms representing intensity modulation and camera shutter speed operation used in encoding / decoding information algorithms, according to some disclosed embodiments. [Figure 26B] 10 illustrates exemplary waveforms representing intensity modulation and camera shutter speed operation used in encoding / decoding information algorithms, according to some disclosed embodiments. [Figure 26C] 10 illustrates exemplary waveforms representing intensity modulation and camera shutter speed operation used in encoding / decoding information algorithms, according to some disclosed embodiments.

[0060] [Figure 27A] 1 illustrates an exemplary modulation scheme for data transmission using a light source according to some disclosed embodiments. [Figure 27B] 1 illustrates an exemplary modulation scheme for data transmission using a light source according to some disclosed embodiments. [Figure 27C] 1 illustrates an exemplary modulation scheme for data transmission using a light source according to some disclosed embodiments.

[0061] [Figure 28]1 is a flowchart illustrating an exemplary method for data association and synthesis, consistent with some disclosed embodiments.

[0062] [Figure 29] 1 is a flowchart illustrating an exemplary method for data association and synthesis, consistent with some disclosed embodiments.

[0063] [Figure 30] FIG. 1 is a schematic diagram illustrating an example arrangement of a line of light in a constellation of light sources, according to some disclosed embodiments.

[0064] [Figure 31A] FIG. 2 is a schematic diagram of an exemplary data encoding scheme consistent with some disclosed embodiments.

[0065] [Figure 31B] FIG. 1 is a schematic diagram of an exemplary encoding scheme for data transmission using a combination of line and point light sources in accordance with some disclosed embodiments.

[0066] [Figure 31C] 1 illustrates a flowchart of an example method for pose estimation using a line light source, consistent with some disclosed embodiments.

[0067] [Figure 32A] FIG. 1 is a schematic diagram of an exemplary GPS multipath error, consistent with some disclosed embodiments. [Figure 32B] FIG. 1 is a schematic diagram of an exemplary GPS multipath error, consistent with some disclosed embodiments.

[0068] [Figure 33] FIG. 1 is a schematic diagram of an example pipeline for data augmentation configured to augment a precision landing and takeoff (PLaTO) system using GPS, in accordance with some disclosed embodiments.

[0069] [Figure 34] FIG. 1 is a schematic diagram of an example pipeline for data augmentation configured to augment GPS using a PLaTO system, in accordance with some disclosed embodiments.

[0070] [Figure 35] FIG. 1 is a schematic diagram of an example pipeline for data augmentation configured to augment an INS using a PLaTO system, in accordance with some disclosed embodiments.

[0071] [Figure 36] FIG. 1 is a schematic diagram of an example pipeline for data augmentation configured to augment an INS with an optical localization system using an Extended Kalman Filter (EKF), in accordance with some disclosed embodiments.

[0072] [Figure 37] 1 illustrates an exemplary system illustrating the integration of a PLaTO system with an aircraft to support piloted or unmanned flight, according to some disclosed embodiments.

[0073] [Figure 38] 10 shows a data plot illustrating aircraft altitude as a function of horizontal distance when the system is used to augment an INS, according to some disclosed embodiments.

[0074] [Figure 39] 10 shows a data plot illustrating ground altitude versus time and corresponding error when the system is used to extend an INS, according to some disclosed embodiments.

[0075] [Figure 40] 10 shows a data plot illustrating easting estimates versus time and corresponding errors when the system is used to extend an INS, according to some disclosed embodiments.

[0076] [Figure 41] 10 shows a data plot illustrating north estimates versus time and corresponding errors when the system is used to augment an INS, according to some disclosed embodiments.

[0077] [Figure 42] 1 is a flowchart illustrating an example method for determining the position and orientation of an airborne vehicle, in accordance with some disclosed embodiments.

[0078] [Figure 43A] 1 illustrates an example of a rapidly deployable constellation of light sources in accordance with some disclosed embodiments. [Figure 43B] 1 illustrates an example of a rapidly deployable constellation of light sources in accordance with some disclosed embodiments.

[0079] [Figure 44] 1 illustrates an example of a VTOL aircraft in accordance with disclosed embodiments.

[0080] [Figure 45] 1 illustrates an example of a VTOL aircraft in accordance with disclosed embodiments.

[0081] [Figure 46] 1 illustrates an exemplary top view of a VTOL aircraft according to disclosed embodiments.

[0082] [Figure 47] 1 illustrates an exemplary propeller rotation for a VTOL aircraft in accordance with disclosed embodiments.

[0083] [Figure 48] 1 illustrates an exemplary power connection for a VTOL aircraft according to disclosed embodiments.

[0084] [Figure 49] 1 illustrates an exemplary architecture of an electric propulsion unit according to disclosed embodiments.

[0085] [Figure 50] 1 illustrates an exemplary top view of a VTOL aircraft according to disclosed embodiments.

[0086] [Figure 51] 10 shows data plots of X, Y, and Z position error as a function of distance from a constellation of light sources obtained from a live data set in accordance with disclosed embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0087] The present disclosure relates primarily to components of electric vertical take-off and landing (eVTOL) aircraft used in non-traditional aircraft. For example, an eVTOL aircraft of the present disclosure may be intended for frequent (e.g., 50 or more flights per workday) of short duration (e.g., less than 100 miles per flight) flights over and into densely populated areas. The aircraft may be intended to carry four to six passengers or commuters who expect a low-noise and low-vibration experience. Therefore, it may be desirable for these components to be configured and designed for frequent use without wear, to generate low heat and vibration, and for the aircraft to include mechanisms to effectively control and manage the heat and vibration generated by the components. Also, several of these aircraft may be intended to operate close to each other in congested urban areas. Therefore, it may be desirable for these components to be configured and designed to generate low levels of noise both inside and outside the aircraft and to include various safety and backup mechanisms. For example, for safety reasons, it may be desirable for the aircraft to be propelled by a distributed propulsion system, avoiding the risk of a single point of failure, and capable of conventional takeoff and landing on runways. Additionally, it may be desirable for an aircraft to be able to safely take off and land vertically from a relatively confined space (e.g., a vertiport, parking lot, or driveway) compared to a traditional airport runway, and to be able to transport four to six passengers or commuters and associated luggage. These usage requirements may impose design constraints on the aircraft's size, weight, and operational efficiency (e.g., drag, energy use), which may affect the design and configuration of aircraft components.

[0088] The disclosed embodiments provide new and improved aircraft component configurations not observed in conventional aircraft and / or identified design criteria that differ from conventional aircraft components. Such alternative configurations and design criteria are combined to address shortcomings and challenges of conventional components, resulting in the embodiments disclosed herein for eVTOL aircraft components of various configurations and designs.

[0089] In some embodiments, the disclosed eVTOL aircraft may be designed for both vertical and conventional takeoff and landing, enabling vertical flight, forward flight, and transitions via a distributed electric propulsion system. Thrust may be generated by supplying high-voltage power to the distributed electric propulsion system's electric engines, each of which can convert the high-voltage power into mechanical shaft power to rotate a propeller. The embodiments disclosed herein may involve optimizing the energy density of the electric propulsion system. The embodiments may include an electric engine connected to an onboard power source, which may include devices capable of storing energy, such as batteries or capacitors, and may also include one or more systems for harnessing or generating electricity, such as a fuel-powered generator or a solar panel array. Some disclosed embodiments provide weight and space savings for aircraft components, improving aircraft efficiency and performance. Focusing on passenger transport safety, the disclosed embodiments implement new and improved safety protocols and system redundancies in the event of a malfunction, minimizing single points of failure in the aircraft propulsion system. Some disclosed embodiments also provide new and improved approaches to meeting aviation and transportation laws and regulations. For example, the Federal Aviation Administration enforces federal laws and regulations that require safety components, such as fire protection barriers, next to engines that use more than a certain amount of oil or other flammable materials.

[0090] In a preferred embodiment, the distributed electric propulsion system may include 12 electric engines that may be mounted on forward and aft booms on the aircraft's wings. The forward electric engines may be tiltable during flight between a horizontally positioned position (e.g., to generate forward thrust) and a vertically positioned position (e.g., to generate vertical lift). The forward electric engines may be either clockwise or counterclockwise in terms of the direction of propeller rotation. The aft electric engines may be fixed in a vertically positioned position (e.g., to generate vertical lift) and may also be clockwise or counterclockwise in terms of the direction of propeller rotation. In some embodiments, the aircraft may have various combinations of forward and aft electric engines. For example, the aircraft may have six forward and six aft electric engines, four forward and four aft electric engines, or any combination of forward and aft engines, including embodiments in which the number of forward and aft electric engines is unequal. In some embodiments, the aircraft has four forward and four aft propellers, at least four of which include tiltable propellers.

[0091] In a preferred embodiment, for vertical take-off and landing (VTOL) missions, the forward and aft electric engines can provide vertical thrust during takeoff and landing. During flight phases in which the aircraft is in forward flight mode, the forward electric engine can provide horizontal thrust, and the propellers of the aft electric engines can be retracted in a fixed position to minimize drag. The aft electric engines can be actively retracted with position monitoring. Transition from vertical to horizontal flight, and vice versa, can be achieved via a tilt propeller subsystem. The tilt propeller subsystem can redistribute thrust from a primarily vertical direction during vertical flight mode to a primarily horizontal direction during forward flight mode. A variable pitch mechanism can change the collective angle of the blades of the propeller hub assembly of the forward electric engine for operation during hover, transition, and cruise phases.

[0092] In some embodiments, during conventional takeoff and landing (CTOL) missions, the forward electric engine may provide horizontal thrust for wing-dependent takeoff, cruise, and landing. In some embodiments, the aft electric engine may not be used to generate thrust during CTOL missions, and the aft propeller may be stowed in place.

[0093] Exemplary embodiments are described herein with reference to the accompanying drawings. The drawings are not necessarily drawn to scale. Examples and features of the principles disclosed herein are described, but modifications, adaptations, and other implementations are possible without departing from the spirit or scope of the disclosed embodiments. Additionally, the words "comprising," "having," "containing," "including," and other similar forms are intended to be semantically equivalent and open-ended, and an item following any of these words does not imply an exhaustive list of the items, nor is it limited to the listed items. Please note that as used in this specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise.

[0094] Throughout this disclosure, references may be made to "disclosed embodiments," which refer to examples of the inventive ideas, concepts, and / or manifestations described herein. Related and unrelated embodiments are described throughout this disclosure. The fact that some "disclosed embodiments" are described as exhibiting a feature or characteristic does not necessarily mean that other disclosed embodiments share that feature or characteristic.

[0095] The embodiments described herein include computer-readable media (e.g., non-transitory computer-readable media) containing instructions that, when executed by at least one processor, cause the at least one processor to perform a method or set of operations. A non-transitory computer-readable medium may be any medium capable of storing data in any memory that can be read by any computing device having a processor to execute the method or other instructions stored in the memory. A non-transitory computer-readable medium may be implemented to include any combination of software, firmware, and hardware. Software may desirably be implemented as an application program tangibly embodied in a program storage unit or computer-readable medium consisting of a component, a particular device, or a combination of devices. The application program may be uploaded to and executed by a machine having any suitable architecture. Preferably, the machine may be implemented on a computer platform having hardware such as one or more central processing units (“CPUs”), memory, and input / output interfaces. The computer platform may include an operating system and / or microinstruction code. Various processes and functions described in this disclosure may be part of the microinstruction code, part of the application program, or a combination thereof, and may be executed by a CPU, whether or not a computer or processor is explicitly indicated. In addition, various peripheral devices may be connected to the computer platform such as an additional data storage unit and a printing unit. Further, the non-transitory computer-readable medium may be any computer-readable medium except a transitory propagating signal.

[0096] Memory may include any mechanism for storing electronic data or instructions, including random access memory (RAM), read-only memory (ROM), hard disk, optical disk, magnetic media, flash memory, or other persistent, fixed, volatile, or non-volatile memory. Memory may include one or more separate storage devices, either uniform or distributed, that can store data structures, instructions, or other data. Memory may also include a memory portion that contains instructions for a processor to execute. Memory may be used as a working memory device for a processor or as temporary storage.

[0097] Some embodiments may include at least one processor. "At least one processor" may constitute any physical device or group of devices having electrical circuitry that performs logical operations on input(s). For example, the at least one processor may include one or more integrated circuits (ICs) that include an application-specific integrated circuit (ASIC), a microchip, a microcontroller, a microprocessor, all or part of a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a field-programmable gate array (FPGA), a server, a virtual server, or other circuitry suitable for executing instructions or performing logical operations. The instructions executed by the at least one processor may be pre-loaded into a memory integrated into or embedded in the controller, for example, or stored in a separate memory.

[0098] In some embodiments, the at least one processor may include multiple processors. Each processor may have a similar structure, or the processors may have different structures that are electrically connected or disconnected from one another. For example, the processors may be separate circuits or integrated into a single circuit. When multiple processors are used, the processors may be configured to operate independently or to operate cooperatively. The processors may be coupled electrically, magnetically, optically, acoustically, mechanically, or by other means that enable interaction.

[0099] As used herein, unless otherwise specified, the term "or" includes all possible combinations unless impracticable. For example, if a component is described as being able to include A or B, then the component can include A, or B, or A and B, unless otherwise specified or impracticable. As a second example, if a component is described as being able to include A, B, or C, then the component can include A, or B, or C, or A and B, or A and C, or B and C, or A, B, and C, unless otherwise specified or impracticable.

[0100] In the following description, various examples are provided for illustrative purposes. However, it should be understood that the present disclosure can be practiced without one or more of these details. Reference will now be made in detail to non-limiting examples of the present disclosure, examples of which are illustrated in the accompanying drawings. The examples are described below with reference to the drawings, in which like reference numerals refer to like elements. When similar reference numerals are shown, the corresponding description(s) will not be repeated and the interested reader is referred to the previously discussed figure(s) for a description of the like element(s).

[0101] Various embodiments are described herein with respect to systems, methods, devices, or computer-readable media. It is intended that one disclosure is a full disclosure. For example, it should be understood that the disclosure of a computer-readable medium described herein also constitutes a disclosure of methods implemented by the medium, and systems or devices for implementing those methods, e.g., via at least one processor. It should be understood that this is a form of disclosure for ease of discussion, and that one or more aspects of one embodiment described herein may be combined with one or more aspects of other embodiments described herein within the intended scope of this disclosure.

[0102] In accordance with this disclosure, some implementations may include a network. The network may comprise any combination or type of physical and / or wireless computer networking arrangement used to exchange data. For example, the network may be the Internet, a private data network, a virtual private network using a public network, a Wi-Fi network, a mesh network, a local area network (LAN), a wide area network (WAN), and / or any other suitable connection or combination that may enable information exchange between various components of the system. In some implementations, the network may include one or more physical links used to exchange data, such as Ethernet, coaxial cable, twisted pair cable, fiber optics, or other suitable physical media for data exchange. The network may include a public wired network or a wireless cellular network. The network may be a secure or non-secure network. In other embodiments, one or more components of the system may communicate directly over a dedicated communications network. The direct communication may use any suitable technology, including BLUETOOTH™, BLUETOOTHLE™ (BLE), Wi-Fi, near field communications (NFC), or any other suitable communications method that provides a medium for exchanging data and / or information between separate entities.

[0103] Reference will now be made in detail to example embodiments illustrated in the accompanying drawings. The following description refers to the accompanying drawings, in which like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments illustrated in the following description of exemplary embodiments do not represent all embodiments consistent with the disclosure. Rather, they are merely examples of apparatus and methods consistent with aspects related to the subject matter described in the accompanying sections. Without limiting the scope of the present disclosure, some embodiments may be described in the context of providing systems and methods for electric vertical take-off and landing (eVTOL) aircraft or airborne vehicles. However, the present disclosure is not so limited. Other types of airborne vehicles, including, but not limited to, unmanned aerial vehicles (UAVs), manned aircraft, conventional vertical take-off and landing (VTOL) aircraft, hybrid VTOLs, and the like, among other airborne vehicles, may utilize the systems and methods disclosed herein.

[0104] Advanced air mobility (AAM) is an emerging field of aviation that involves the use of small aircraft for daily transportation and other services. Many AAM aircraft are expected to take off and land at new infrastructure called vertiports. As described herein, a vertiport refers to a landing site or surface for an aerial vehicle, such as an eVTOL, to land or take off. In some embodiments, a vertiport may also be referred to as a vertiplex or vertistop. The location of a vertiport may be determined based on various factors, including, but not limited to, physical obstructions, federal and state or local regulatory restrictions, surrounding uses, etc. Physical obstructions may be fixed, anticipated, movable, or temporary. An example of an anticipated physical obstruction may include an adjacent lot with development rights for a 40-story building that is currently vacant. Examples of physical obstructions may include nearby high-rise buildings, antennas, towers (cell and water), trees, power lines, utility poles, signs, the land use designation of the vertiport site, owner rights, etc.

[0105] In some cases, regulatory restrictions may include both the vertiport site's current land-use designation and the owner's rights. For example, in an air rights transaction or transfer of development rights, an owner may sell the right to build above their property to a buyer who wants to build something larger than would normally be permitted. For example, if a parking lot operator sells the air rights above their parking lot, a proposed vertiport terminal extending into that space likely cannot be built without the air rights owner's approval. Height districts are geographic areas where maximum building heights are restricted, and these should also be considered when considering vertiport siting. Physical considerations, including physical barriers, can be evaluated and balanced, taking into account expected future development patterns and the jurisdiction's vision, in an attempt to accommodate development such as population shifts, increased or decreased density, and the current trend toward mixed-use neighborhoods with residential and commercial buildings in close proximity.

[0106] Mobile or temporary physical obstructions include structures that are variable or temporary in nature. Mobile or temporary physical obstructions may include both planned and anticipated considerations. Planned considerations include a process by which vertiport operators have an opportunity to provide input, while anticipated considerations are those that occur with little or no advance notice but are likely to occur during the vertiport's operational life. Some examples of temporary structures may include temporary vertiports, construction cranes, blown debris, construction staging, noise, lightning protection devices, non-acoustic nuisance factors, electrostatic discharge, urban wind shadow, or future local land use. These considerations reflect temporary and potentially insignificant events during the vertiport's operational life but are still worth considering to support safe and efficient operations. Additionally, vertiport siting decisions may also be influenced by the expected frequency of certain temporary considerations. For example, locating a vertiport next to a tall tree increases the likelihood that debris will periodically enter the vertiport's operating area in the future as the tree grows, potentially encroaching on the vertiport's airspace and posing a hazard to navigable airspace.

[0107] When selecting a vertiport location and designing vertiport operations, it can be important to consider the surrounding area. Surrounding uses include considerations that occur outside the vertiport site but within the local neighborhood. These considerations may affect the vertiport during site selection, design, and operation, and may change over the vertiport's lifespan. A vertiport may also affect the surrounding area and modify these considerations. Examples of surrounding uses that may affect vertiport site selection include critical infrastructure, local fire stations, subway and bus stops, local land use, distance to maintenance and repair facilities, and downwind of wind farms. In some cases, surrounding uses may be affected by the vertiport. Examples of this scenario include nearby schools, properties on approach and takeoff paths, noise-sensitive areas, visual disturbances (e.g., solar panel reflections), zoos, protected wildlife habitats, and the privacy of vertiport neighbors. The proximity of a vertiport to existing infrastructure can be a major location factor. Infrastructure considerations include current local land uses (e.g., schools, hospitals, parks, or other noise-sensitive areas), emergency response (e.g., fire stations), and direct connections to other transportation options (i.e., intermodality). For initial vertiport locations, proximity to these types of existing infrastructure allows for rapid development and operation by shortening the development lead time for ancillary criteria (such as land uses designated for transportation). On the other hand, vertiports located too close to other types of infrastructure may impede flight operations. For example, proximity to wind farms may restrict approach and takeoff paths and create airflow disturbances that may impede safe flight operations. Vertiport design and configuration may include several other factors, including, but not limited to, economic considerations such as aircraft performance in the vertiport environment, passenger comfort, development costs, maintenance costs, and revenue generation; environmental considerations; airspace considerations; demand considerations; contingency considerations; communications and data management; security considerations; safety and utility; and automation.

[0108] In urban air mobility applications, takeoff and landing of eVTOL aircraft in urban environments may require highly accurate and reliable positioning capable of operating in GNSS-challenged environments. As used herein, a GPS-unavailable or GPS-challenged environment refers to an environment lacking reliable access to Global Positioning System (GPS) or Global Navigation Satellite System (GNSS) signals. In GPS-unavailable environments, GPS signals may be degraded, interrupted, denied, jammed, hacked, or simply rendered ineffective by multipath effects or satellite signal obstruction. Satellite signals may be rejected in challenging environments due to the lack of a clear line-of-sight path between the satellite and the user antenna. Signals may be interrupted or degraded by adverse weather conditions, poor visibility, high-rise buildings in urban areas, and hostile or uncooperative landing conditions, among other factors.

[0109] An air vehicle (e.g., air vehicle 310 in FIG. 3 ) may include an electric air vehicle or eVTOL vehicle. eVTOL vehicles may be used for on-demand urban air transportation services, potentially providing alternative transportation in urban environments with low direct operating costs, low noise, and zero emissions. Furthermore, eVTOL aircraft offer an alternative air transportation that is versatile (capable of vertically taking off and landing from basic landing zones), economical (reduced acquisition and operating costs), accessible (usable by operators with little or no flight training or experience), and safe (designed for improved tolerance to failure modes). In this regard, the development of distributed electric propulsion (DEP) may enable inexpensive, quiet, and reliable short-range VTOL aircraft. The use of DEP may provide significant flexibility, enabling new aircraft configurations, architectures, and control methods. Furthermore, electric propulsion is scalable in that very similar levels of motor power, weight, and efficiency can be achieved over a wide scaling range. For example, redundant DEPs can be used to improve in-flight resilience and safety. The use of electric motors can improve safety on the ground through reduced noise, heat dissipation, and possible toxic fumes, and it also allows for the stopping of rotating propellers and rotors before passengers board or disembark. Furthermore, while DEP propellers and ducted fans still generate noise (level and frequency dependent on tip speed, disk load, and other design parameters), the combination of DEP configuration (multiple small rotors directly driven by electric motors), tip speed limitation, elimination or minimization of engine / turbine / gear noise sources, and the potential use of fixed wings for efficient forward flight means that eVTOLs are expected to have modified low-noise characteristics compared to similarly sized conventional helicopters, with target noise reductions of 15 dB or more. Battery-powered eVTOL aircraft could also reduce their environmental impact with zero operational emissions. Furthermore, the use of DEPs instead of complex shafts, cross-couplings, and gear arrangements is expected to reduce both acquisition, maintenance, and operating costs. If extended range is required, aircraft can be designed with a hybrid electric propulsion system, taking advantage of the benefits of operating small engines at peak efficiency.Operating the hybrid power unit engine at idle or off during takeoff and landing can further reduce the noise signature of the aircraft at low altitudes.

[0110] FIG. 1A shows a schematic of a conventional instrument landing system (ILS) that provides horizontal and vertical guidance to guide an aircraft along a runway. The ILS is a standard precision landing aid used to provide precise azimuth and descent guidance signals to guide an aircraft to land on a runway under normal or adverse weather conditions. An ILS may include three subsystems: a localizer, a glideslope, and a marker beacon. The localizer provides horizontal guidance to approaching aircraft, as shown in FIG. 1B. The glideslope provides vertical guidance to approaching aircraft, as shown in FIG. 1C. The marker beacon provides distance information as the approach progresses. In some cases, the marker beacon may be replaced by a distance measuring device (DME). Additionally, an ILS may include high-intensity lighting at the end of the runway to help pilots find the runway and transition from approach to visual landing.

[0111] In an ILS system, two or more radio frequencies (RF) are broadcast, one spatially offset from the other, as shown in Figures 1B and 1C. The spatially offset RF signals are replicated both horizontally and vertically. In some cases, signal sensors associated with the aircraft may measure the strength of the two signals. If one signal is stronger than the other, the aircraft will deviate from its planned landing trajectory. Based on this information, the pilot can constantly correct course to align with the centerline and utilize a glideslope to the runway. While ILS and related systems are established for conventional passenger aircraft, they may not be suitable for eVTOL aircraft in urban environments due to non-standard flight approach trajectories, interference with nearby ILS systems and all buildings, or the large footprint of an ILS.

[0112] Reference is now made to FIG. 2 , which illustrates a schematic diagram of an exemplary landing / takeoff approach using optical navigation in a GPS-disabled environment, in accordance with some embodiments of the present disclosure. An example view 200 of an eVTOL aircraft with an active marker in the camera's field of view approaching a landing site or landing area in a GPS-disabled environment, e.g., an urban area with dense high-rise buildings, is shown in FIG. 2 . An exemplary airborne vehicle 210, such as an eVTOL, can approach a vertiport 220 using one or more approach paths 260. One of several advantages of using an eVTOL is that the airborne vehicle can approach a vertiport from any direction, unlike the corridor-based landing approach of a traditional ILS system. By way of example, view 200 illustrates another approach direction (approach 2). In some embodiments, the approach angle may vary between 7° and 9°, as illustrated. The airborne vehicle 210 may include a camera (not shown in FIG. 2 but described in a later section) with a field of view 270.

[0113] An exemplary precision landing and takeoff system and data communications system according to some embodiments of the present disclosure is now illustrated. The precision landing and takeoff system described herein refers to an optical navigation-based eVTOL positioning system that operates in GPS-unavailable environments. See FIG. 3 . As previously mentioned, eVTOL airborne vehicles, such as airborne vehicle 310, may be used in urban air mobility (UAM) applications that generate commercial passenger services such as air taxis, as well as public service applications such as firefighting, delivery of medical assistance, emergency search and rescue operations, disaster relief operations, and law enforcement. In some embodiments, the aircraft may be autonomous, i.e., operated without a pilot.

[0114] The precision landing and takeoff system may include an airborne vehicle 310 with an onboard optical detection device 315, a vertiport 320 including a marker 350, and a ground control unit 330. The optical detection device 315 may include a camera 311 and a processor 312. In some embodiments, the optical detection device 315, the ground control unit 330, and the one or more vertiports 320 may communicate wirelessly with each other during landing or takeoff operations of the eVTOL airborne vehicle 310. Communication between the optical detection device 315, the ground control unit 330, and the one or more vertiports 320 may include receiving and transmitting data and information related to providing landing or takeoff guidance to the airborne vehicle 310.

[0115] The precision landing and takeoff system may include one or more vertiports 320 (also shown as vertiports 420 in FIG. 4). In some embodiments, vertiport 320 or vertiport 420 may comprise a landing surface or landing location for an eVTOL vehicle, such as airborne vehicle 310. Vertiport 320 may include multiple light sources arranged in a predetermined pattern, with the characteristics of the light emitted from each light source configured to be modulated over time. Each vertiport may include an active constellation of markers 350 or active light sources. As used herein, an active light source (ALS) refers to a light source whose characteristics are modulated over time. For example, the intensity of light emitted from an active light source may be modulated over time. Other characteristics that may be modulated over time include, but are not limited to, the frequency, amplitude, wavelength, phase, bandwidth, or duty cycle of the emitted light. Examples of active light sources may include, but are not limited to, light emitting diodes (LEDs).

[0116] In some embodiments, the landing and takeoff area of ​​vertiport 320 may be rectangular, circular, triangular, substantially rectangular, substantially circular, or substantially triangular, or a combination thereof, or any other suitable shape. As used herein, landing and takeoff area refers to an area of ​​a vertiport where an airborne vehicle (e.g., airborne vehicle 310) can land or takeoff. In some embodiments, the predetermined pattern in which the active light sources are arranged may resemble the shape of the landing and takeoff area, such that the active light sources define the boundaries of the landing and takeoff area. In some embodiments, the predetermined pattern of active light sources defining the boundaries of the landing and takeoff area may include low-intensity light sources that enter the field of view of a camera to assist in landing or takeoff as the airborne vehicle approaches the landing target.

[0117] In some embodiments, the active light sources may be arranged in a substantially axisymmetric shape, such as a circle, square, or rectangle. In some embodiments, the active light sources may be arranged in an elliptical, triangular, trapezoidal, or other shape. In some embodiments, the active light sources may be evenly or unevenly spaced in an axisymmetric shape. In an unevenly spaced arrangement in an axisymmetric shape, the distance between adjacent active light sources may not be uniform. In some embodiments, the active light sources may be arranged in a grid-based pattern that is evenly spaced across the landing platform. Other suitable arrangements are possible.

[0118] In some embodiments, the active light sources may be arranged in an asymmetrical shape to maximize the detectability, uniqueness, or spoof and jamming resistance of the vertiport with which they are associated. An example of an asymmetrical shape for the arrangement of active light sources is shown in Figure 5 (discussed later).

[0119] In some embodiments, markers 350 may include a constellation of fiducial light sources (e.g., active light sources) in the infrared (IR) or visible spectrum distributed at known locations on vertiport 320. In this regard, the precision landing and takeoff system may be referred to as an Active Fiducial Light Pattern Localization (AFLPL) system. Some advantages of using active light sources as markers at vertiports for optical navigation of eVTOL airborne vehicles include the following: i. Authentication and Security - One or more properties of the emitted light can be modulated to enable authentication and improve security. ii. be easily distinguishable from the surrounding environment, allowing for the reduction or elimination of visual clutter in urban environments; iii. Enhanced Range, Detection Capability, Data Transmission - One or more characteristics may be modulated to enhance range, detection, data transmission, etc., to convey messages to aircraft or enable low bandwidth communications. iv. Day / Night Capability - Airborne vehicles can operate both during the day and at night. v. Jamming and Spoofing Resistance - Optically based AFLPL approaches are inherently more robust to jamming and spoofing than radio frequency (RF) signals. The modulation capabilities of AFLPL may further enable the implementation of authentication codes. vi. No Significant Regulatory Obstacles - The AFLPL approach may offer an easier path to certification by the FAA, increasing the acceptability and feasibility of implementing active light sources in public service applications. Furthermore, unlike RF-based ranging and communication approaches, AFLPL does not require FCC allocation of the limited available RF spectrum. Furthermore, in military applications, AFLPL's narrowband near-infrared illumination may provide lower detection levels than RF radiation. vii. High Positioning and Guidance Accuracy - Positioning accuracy compared to GPS or GNSS may be equivalent or better near the vertiport, especially in environments where GPS is difficult or unavailable. Position accuracy error may be less than 1 m within 100 m of the vertiport and decreases as the vehicle approaches the vertiport. In some cases, position accuracy error may be less than 10 cm within 10 m of the vertiport. viii. High redundancy and high reliability - Active light sources can be a relatively low cost option and because they have no weight in the air, multiple light sources can be used at the vertiport to provide uninterrupted guidance. ix. Enhanced Visibility - Infrared (IR) wavelengths of the emitted light may enhance visibility and detectability in adverse weather conditions such as fog, rain, storms, and lightning. x. Low Implementation Complexity - A vertiport may require minimal infrastructure. Visible and infrared light sources can be integrated into the vertiport structure and embedded into the landing pad with relatively low technical complexity and cost.

[0120] In some embodiments, the location of the active light sources in a constellation within the vertiport may be designed to provide optimized localization throughout the landing trajectory. In some embodiments, the constellation may include a first set of light sources arranged in a first predetermined pattern, each of the light sources in the first set configured to be located within the field of view of a camera associated with the airborne vehicle when the airborne vehicle is a first distance from the landing surface. The first set of light sources may include light sources having higher intensity and located at a greater distance from the landing target, which may improve performance when the airborne vehicle is at a longer distance from the vertiport. The light sources in the first set may be within the field of view of the camera at a greater distance and may be outside the field of view of the camera when the airborne vehicle is closer to the vertiport or the landing target. The constellation may further include a second set of light sources arranged in a second predetermined pattern, each of the light sources in the second set configured to be located within the field of view of the camera when the airborne vehicle is a second distance from the landing surface. The second set of light sources may include light sources with lower intensity and located within a smaller distance, such that the second set of light sources remain within the camera's field of view when the airborne vehicle is final approach or within a predetermined approach distance. The intensity of the second set of light sources may be different from the intensity of the first set of light sources so as not to interfere with detection of the first set of light sources from a greater distance. In some embodiments, the intensity of the first set of light sources may be higher than the intensity of the second set of light sources. In some embodiments, the area covered by the first set of light sources may be larger than the area covered by the second set of light sources. In other words, the first set of light sources may be distributed over a wider area compared to the second set of light sources so that only the second set of light sources can be detected when the airborne vehicle is within a predetermined approach distance.

[0121] In some embodiments, the light sources in the constellation may be arranged to maximize the detectability of each location by maximizing the spacing between each light source. In some embodiments, the light sources may be arranged to maximize the ability to identify a light source from multiple sources, such as by minimizing the symmetry of the arrangement pattern. Additionally, in some embodiments, a predetermined pattern of light sources may be associated with a landing surface. For example, the constellation may include a uniquely identifiable pattern for each vertiport, such that the vertiport may be identified based on the arrangement pattern of the light sources within the vertiport.

[0122] In some embodiments, active light sources (e.g., fiducials) may additionally or alternatively be placed on the ground, on buildings, or on other objects along the common flight path. Possible locations include, but are not limited to, on roads, streetlights, buildings, antennas, and other tall structures. This allows for accurate position information to be obtained during flight, not just when the landing surface or landing pad is in view. Furthermore, because the same algorithms can be used to calculate position, but the light source pattern is spread over a much wider area, the system may function throughout the entire or majority of the flight, not just during final approach to the landing pad.

[0123] In some embodiments, one or more vertiports may have different sizes and dimensions. The size of the vertiport may determine the degree of performance when the airborne vehicle is farther from the vertiport. The fiducial pattern within the constellation may not need to be the same at all vertiports, although prior knowledge of the layout or placement of the fiducial light markers may be desirable. However, in some embodiments, the light sources may be configured to transmit information related to their location, in which case prior knowledge of the light source's location or placement may not be required. In some embodiments, the constellation pattern or placement may be stored in a system database or server. Information stored in the database may be accessible and updatable in real time or based on input from a user.

[0124] In some embodiments, one or more characteristics of the emitted light may be modulated to convey information. Modulation may be performed by one or more methods, including, but not limited to, switching on or off, frequency modulation, amplitude modulation, duty cycle modulation, synchronization options, etc. The conveyed information may include the identity of the vertiport, the location of the light source, the identity of the light source, and the operational status of the vertiport. In some embodiments, the conveyed information may include an encoded signal authenticating the landing surface or vertiport. It should be understood that a combination of modulation methods and the information conveyed by modulation may be applied as appropriate. For example, the emission frequency of the light emission from the light source may be modulated to convey information related to the identity of the light source, and the duty cycle may be modulated to convey information related to the identity of the vertiport. In another example, the emission frequency of the light emission from the light source may be modulated to convey information related to the identity of the vertiport. In some embodiments, the frequency may be modulated to indicate the operational status of the vertiport, such as functioning, not functioning, under maintenance, etc.

[0125] In some embodiments, the wavelength of light emitted from one or more light sources may be determined based on several factors, including, but not limited to, maximizing the difference between the emitted light and background light to improve detectability, minimizing absorption of the emitted light by atmospheric or weather effects, maximizing the sensitivity of detection by a camera, or reducing visible light contamination around the landing target, among others.

[0126] In some embodiments, the wavelength of light emitted from one or more light sources is in the range of 800-1550 nm. In some embodiments, the wavelength of light emitted from one or more light sources is in the range of 800-850 nm. In preferred embodiments, the wavelength of light emitted from one or more light sources is 810 nm. In some embodiments, the wavelength of light emitted from one or more light sources is 1310 nm. In some embodiments, the wavelength of light emitted from one or more light sources is 1550 nm.

[0127] In some embodiments, one or more light sources 450 may be recessed relative to the landing surface of the vertiport 420, as shown in FIG. 4A. An exemplary light source 450 is shown in FIG. 4B. In some embodiments, one or more light sources may be recessed, protruding, or flush with the landing surface. In some embodiments, the light source 450 may include a protective or encapsulating cover to prevent the ingress of moisture, dust, or other particles that may adversely affect the performance of the light source. In some embodiments, the protective cover may be configured to transmit substantially all of the light emitted from the light source, resulting in little or no absorption by the protective cover.

[0128] In some embodiments, each light source may further include an optical sensor configured to detect a portion of the light emitted from at least one other of the light sources. This may be desirable to synchronize the camera capture rate with the modulation of the light sources to reduce errors that may occur when images are captured during transitions (discussed in a later section). In some embodiments, the landing surface or vertiport may further include controller circuitry configured to operate the light sources. In this context, operating a light source may include activating, deactivating, or modulating characteristics of the light source, for example, by adjusting an electrical signal applied to the corresponding light source. The controller circuitry may further include time management circuitry, power management circuitry, sequencing circuitry, etc. In some embodiments, one or more processors may be configured to remotely control the operation of the light sources.

[0129] In some embodiments, the wavelength of the emitted light can be modulated to increase the detectability or range of the light source or to transmit data from a corresponding fiducial. The inventors recognize that while the wavelength can be tuned, this can negatively impact the simplicity and implementability of the system. For example, to detect wavelength variations, a hyperspectral camera may be required. While such cameras exist and are commercially available, they can be complex, unreliable, and computationally intensive. Furthermore, because the wavelength of an LED is primarily determined by the material composition and transmission medium, such light sources can be experimental, unreliable, and expensive.

[0130] In some embodiments, vertiport 320 (or vertiport 420) may be a portable landing surface. The portable vertiport (not shown here) may include a redeployable landing mat, cloth, or tarp. This may be particularly useful when landing in uncooperative locations where limited or temporary landing infrastructure is required, such as for military operations, firefighting, disaster relief, or medical support operations. In some embodiments, the portable vertiport may include an integrated, battery-powered active light source that can be remotely activated, deactivated, or modulated.

[0131] In some embodiments, the vertiport may include multiple landing surfaces, each of which includes multiple light sources arranged in a predetermined pattern, and the characteristics of the light emitted from each light source configured to be modulated over time.

[0132] In some embodiments, one or more landing surfaces of a vertiport may be horizontally offset from one another, for example, a vertiport hub, a vertiplex, or a large area containing multiple vertiports. Horizontally displaced vertiports may be coplanar or substantially coplanar. As used herein, the terms "coplanar" or "substantially coplanar" refer to the vertiport's landing surface being on the ground, similar to a car parked in a parking lot. In some embodiments, one or more vertiports may be vertically offset from one another so as to be non-coplanar, for example, in a vertiport garage containing multiple levels of vertiports. In some embodiments, one or more vertiports may be horizontally and vertically offset from one another so as to be offset in horizontal and vertical axes, allowing for a higher density of vertiports.

[0133] Returning to FIG. 3 , airborne vehicle 310 may include optical detection device 315 and further includes camera 311 and processor 312. Camera 311 may be configured to generate images based on information transmitted by a light source (e.g., active light source 350) adjacent to the landing surface of the airborne vehicle. For example, camera 311 may be configured to capture images of light emitted from the light source. Processor 312 may include controller circuitry configured to receive the generated images and determine a position and orientation of the airborne vehicle based on the received images.

[0134] In some embodiments, camera 311 may include a color, monochrome, or hyperspectral camera. Camera 311 may be mounted on airborne vehicle 310 such that camera 311 can provide a plan view of the light sources on the landing surface. A plan view of the light sources may be desirable when the airborne vehicle is within landing distance, performing a vertical descent from 50 feet above ground level (AGL) to 0 AGL, or when the airborne vehicle is taking off. In some embodiments, camera 311 may be mounted on airborne vehicle 310 such that camera 311 can provide a forward view of the light sources on the landing surface. This may be desirable to maximize visibility during approach. In some embodiments, one or more cameras may be mounted on the airborne vehicle to capture multiple frames or perspectives from different angles during landing or takeoff.

[0135] In some embodiments, the camera 311 may be always on but is activated to capture images or report measurements upon detecting a fiducial or active light source on the landing surface. Alternatively, the camera 311 may be turned on or activated when the airborne vehicle is within detection distance to conserve power. In another embodiment, the camera 311 may be turned on or activated for a predetermined period of time, by an activation signal from an external processor (e.g., a flight control computer or ground control unit 330), or by the airborne vehicle operator. In some embodiments, the camera 311 may be configured to be activated after the airborne vehicle is within a predetermined distance of the landing surface. The predetermined distance may be based on several factors, including, but not limited to, weather conditions, landing surface conditions, etc. In a preferred embodiment, the predetermined distance may be 500 meters or less.

[0136] In some embodiments, camera 311 may include optical filters configured to permit a range of wavelengths of light emitted from each light source. In other words, the optical filters of camera 311 may be configured to reject wavelengths significantly different from the fiducial transmission wavelength. As used herein, fiducial transmission wavelength refers to the wavelength or wavelength range of light emitted from one or more fiducial markers (e.g., active light sources on a landing surface). For example, if the active light sources are configured to emit light at a wavelength of 810 nm, the optical filter may permit a wavelength range of 808 nm to 812 nm and reject wavelengths outside the permitted transmission range. In some embodiments, the sensitivity of the optical detection of camera 311 can be adjusted to filter incoming wavelengths.

[0137] In some embodiments, the allowed wavelength range is in the range of 800 nm to 850 nm. In a preferred embodiment, the allowed wavelength range is approximately 810 nm. In some embodiments, the allowed wavelength range is approximately 1310 nm. In some embodiments, the allowed wavelength range is approximately 1550 nm. In some embodiments, the optical filter may be configured to allow wavelengths corresponding to the emitted light. As used herein, the term "approximately" refers to an approximation indicating that the allowed wavelength range is within ±2 nm. The optical filter may be a low-pass, high-pass, or band-pass filter. Based on the detected light source, the camera 311 may generate an FPA image as shown in FIG. 5.

[0138] Reference is now made to FIG. 6 , which illustrates an exemplary overview of an operational algorithm and data pipeline 600 of a precision landing and takeoff system according to some embodiments of the present disclosure. As shown, a camera (e.g., camera 311) may be configured to receive light signals from light sources on the vertiport and background information, such as scene information. The camera, including an optical filter, may be configured to output a series of still images, or a video stream, or an FPA image. Based on the generated output signals, a processor may be configured to execute one or more algorithms to detect, associate, and estimate the pose of the airborne vehicle and decode the information in the encoded signal from the active light source. The detection algorithm enables locating the active light source within the image frames captured by the camera. The association algorithm enables associating or mapping the identified active light source within the image to a corresponding active light source on the landing surface. The pose estimation algorithm enables determining the pose of the airborne vehicle based on the associated active light source.

[0139] detection Locating the active light source within an image frame may include distinguishing the received signal from background noise signals. This may be performed using background subtraction and thresholding. If the active light source is modulated so that it is fully on in some frames and fully off in other frames, the frame in which the light source is fully off may be used as a background image to remove the background from the fully on image through subtraction. An exemplary subtraction algorithm is provided here. It should be understood that other suitable subtraction and thresholding techniques may be used to locate the active light source within an image.

[0140] As an example, if I_n are the pixel values ​​of an MxN matrix corresponding to the nth image, then I_diff = abs(I_n - I_(n-1)) is the difference in pixel values ​​between consecutive images. The resulting difference frame (I_diff) may be thresholded to generate a mask that identifies which pixels in the original image correspond to active light sources. Thus, I_mask = (I_diff > threshold). With the active light sources identified in the images, their positions can be calculated through a centroid calculation, which allows the location of the light source to be calculated with sub-pixel accuracy.

number

[0141] In some embodiments, refining the detection or identification of the identified location of an active light source within an image frame may include performing a registration method. This may be desirable for images in which the background is moving rapidly due to camera motion, such as a camera mounted on a moving aerial vehicle. The active light source may change position within the image between on-frame and off-frame. In this case, it may be necessary to shift the images to align them frame by frame in order to subtract the background. To achieve this, any of several techniques may be used, including feature matching, transform matching, or current state estimation. In feature matching, features common to each frame may be identified and their location within each frame may be determined. The images may be transformed and / or warped to align the features. In transform matching, for small changes, the two images may be shifted by one pixel to identify where the background most closely matches. In current state estimation, if the camera position and rotational velocity are known or can be estimated from previous images, the image distortion required to align each frame can be estimated. Furthermore, registration may be performed globally or locally in one or more regions of interest around the active light source. It should be understood that other techniques for improving detection may be used in place of or in combination with the techniques described herein.

[0142] In some embodiments, locating or improving detection of an active light source within an image frame may include tracking the location of the identified active light source to reduce the computational complexity of subsequent calculations. Measurements of velocity and change over time may be useful to properly predict the location of the active light source. This may be obtained by using the aircraft's position and velocity from an external system such as a GPS or inertial navigation system (INS), by using internal estimates of the aircraft's position and velocity derived from changes in position over time, or by using changes in pixel position over time, such as tracking changes in the active light source between frames and extrapolating in time. Tracking may be used to reduce computational complexity by calculating a region of interest for detection rather than the entire image, to improve accuracy by providing estimates for registration, or to calculate additional information such as velocity and acceleration that can be reported to other devices on the aircraft.

[0143] FIG. 7 illustrates an exemplary detection algorithm 700 for locating a light source in an image captured by a camera, in accordance with some embodiments of the present disclosure. As shown, the camera may receive a noisy signal and generate an output, typically in the form of an image or a stream of images. The image may be processed by applying a band-pass filter, performing a Discrete Fourier Transform (DFT) calculation on the filtered image, applying a low-pass filter, and thresholding. FIGS. 8 and 9 illustrate data plots comparing a signal before and after signal processing using a band-pass filter and a DFT calculation, in accordance with some embodiments of the present disclosure. In some embodiments, one or more characteristics of an active light source may be modulated over time to improve detection of the active light source in a captured image.

[0144] Data Association Data association, as used in the context of this disclosure, refers to the process of matching detected points in a camera image to known points in a database of known locations of active light sources, optical markers, or fiducials on the landing surface. Some existing techniques for data association may involve modulating one or more light sources to convey the light source's unique identity and determining location based on the light modulation. However, such approaches can have challenges, including, but not limited to, inaccurate identification due to cross-signaling, low signal-to-noise ratio (SNR), high background noise, etc. As described above and disclosed in several embodiments of this disclosure, one or more data association algorithms can be implemented to map detected points in a camera image to known locations of active light sources. The selection of a data association algorithm may depend on several factors, including, but not limited to, the presence or absence of acceptable data associations, reliability, accuracy, and robustness of the resulting associations. As an example, after finding an acceptable and correct association, a point tracking method (described in more detail below) can be used to track points from image to image. As another example, a grid association algorithm may not rely on an initial association, but the algorithm may not always produce a solution. Therefore, it may be desirable to run two or more data association algorithms in parallel or sequentially to establish acceptable data associations and associate each identified fiducial in a camera image with the identified fiducial's three-dimensional (3D) location. Certain aspects of the present disclosure are directed to data association methods and systems and advantages thereof.

[0145] a. Iterative Closest Point Algorithm In some embodiments, an iterative closest point (ICP) algorithm may be used, which may include the following steps: 1. Given a position estimate, estimate where the ALS is located in the image. 2. For each detected point, find the closest (geometric distance) estimated point. 3. Calculate the sum of the distances between each pair of points (measured vs. estimated). 4. Determine small perturbations (x-y translations and rotations) to the estimated points that will reduce the error distance calculated in step 3. 5. Update the estimated position of the point based on the perturbation calculated in step 4. 6. Return to step 2 until the error converges. 7. The resulting associations are taken from step 2 (nearest neighbors) after the solution has converged.

[0146] Although simple to implement, ICP can be sensitive to inaccuracies in the initial pose estimate: if the step size is too small, it can take a long time to converge and can be very computationally intensive.

[0147] b. Thin-plate spline robust point matching algorithm A spline is a numerical function defined by a polynomial function. Spline functions have a high degree of smoothness at the points where the polynomial parts connect, i.e., at the nodes. Feature-based methods for nonrigid registration may face challenges related to point correspondence between two or more feature sets. In this context, correspondence between two feature sets refers to the association between each identified fiducial in a 2D photograph and the fiducial's 3D location (e.g., an active light source on the ground). The framework of nonrigid point matching or robust point matching (RPM) algorithms can be extended to include spline-based transformations, particularly thin-plate splines. Some methods for solving both correspondences and transformations include ICP (mentioned above). ICP algorithms utilize nearest-neighbor relationships to assign binary correspondences at each step. This correspondence estimate is then used to refine the transformation, and vice versa. While ICP algorithms are simple, fast, and guaranteed to converge to a local minimum, they may be insufficient for nonrigid transformations, especially when deformations are large. Furthermore, outliers rapidly degrade correspondences, making ICP algorithms inappropriate. Generating a smoothly interpolated spatial mapping according to two sets of landmark points is a general problem in spline theory, since there are an infinite number of ways to map one set of points onto another once non-rigidity is allowed. A smoothness constraint is desired to suppress overly arbitrary mappings and outliers. In other words, the behavior of the mapping can be controlled by choosing a particular smoothness factor based on prior knowledge.

[0148] Reference is now made to Figures 13A-13D, which illustrate the correspondence and matching process using a thin-plate spline robust point matching algorithm, according to some disclosed embodiments.

[0149] FIG. 13A illustrates a coordinate space including a normalized grid aligned according to the order of active light sources on the ground. The exemplary normalized grid illustrated in FIG. 13A includes 25 points in a 5×5 array format. Each point on the normalized grid is numbered from 0 to 24 according to the order of the lights on the ground. In some embodiments, the normalized grid may be rotated based on the heading of the airborne vehicle approaching the landing surface. The coordinate space further illustrates normalized detection points received from the detection algorithm. The normalized detection points are indicated by a “+” marker. The normalized detection points may be overlaid with the normalized grid points. The circular boundary indicates the radius of potential matches evaluated in robust point matching.

[0150] The Thin Plate Spline Robust Point Matching (TPS-RPM) algorithm may include performing distance-based point matching and association over a larger search area. Figure 13B shows a coordinate space containing the normalized grid points and the locations of the detected points after the Thin Plate Spline transformation (e.g., as shown in Figure 13D). Lines can be drawn connecting the normalized grid points with the corresponding normalized detected points after the Thin Plate Spline transformation, if any.

[0151] The TPS-RPM algorithm may further include determining TPS distortions that can be used to reduce the error of the associated points. The algorithm may further include reducing the size of the search region and iteratively performing the TPS distortions until the normalized sensed points and normalized grid points converge. Figure 13C shows the association of the normalized sensed points with the associated normalized grid points, indicated by the line connecting the two.

[0152] In some embodiments, given the iterative nature of the TPS-RPM algorithm, using a normalized initial guess to match known patterns to detected points can take a very long time. Therefore, to improve the overall performance of the algorithm, especially when the aircraft approaches a vertiport where there is significant frame-to-frame variation, the TPS-RPM algorithm may be run in two modes. The first mode uses a normalized initial guess to estimate the system's pose. Based on a confidence metric for the estimated pose, other association algorithms, such as ICP, cross-ratio, or grid association, can be used to complement or aid the TPS-RPM by seeding it with the aircraft's current pose, which is then used to project the known pattern onto the camera frame, and the projection can be used as an initial guess before normalization. This method may, in practice, significantly reduce the number of iterations required to converge on a solution, allowing the TPS-RPM to be used at its full rate.

[0153] c. Grid-based association algorithm Reference is now made to FIG. 14A , which illustrates an exemplary camera image 1400 with detected points (shown in a square grid) according to disclosed embodiments. Camera image 1400 may be an image captured by a camera mounted on an airborne vehicle approaching a landing surface at a non-zero angle relative to a plane normal to the landing surface. In some embodiments, image 1400 may include any number of detected points in an array or non-array arrangement. Note that while image 1400 shows a 5×5 grid of detected points, other grid patterns or random patterns including any number of detected points may also be used. In some embodiments, the detected fiducials or points may be numbered in a known, recognizable order. For example, the detected points in image 1400 may be numbered 0 through 24 starting from the top row, moving from top to left and then to right, with the left-to-right numbering repeated for each row. It should be understood that the numbering order should not be construed as limiting in any way. The numbering format is exemplary and not limiting.

[0154] FIG. 14B illustrates a normalized image space 1410 containing normalized detected points for image 1400. In some embodiments, the detected points may be normalized to fit within a predefined image space, e.g., −100 to +100 arbitrary units (au), as shown in FIG. 14B. In some embodiments, normalization may include mean-centering the detected points, constructing a transformation matrix using the mean and variance of the points to make the variance of the detected points equal to 1 (value 1). Some advantages of normalizing the detected points within the normalized image space are that all points are within unit distance and there is no offset between the points. Additionally, processing (e.g., data manipulation) performed on one or more points may affect the remaining points as well.

[0155] After normalizing the detected points, lines in the normalized image space 1410 that pass through multiple detected points can be identified by performing a Hough transform technique, as shown in the normalized image space 1410 of FIG. 14B. The Hough transform technique referred to herein may be used to isolate specific shape features in an image that contain multiple points in parametric form, such as lines, circles, or ellipses. In line detection using the Hough transform, each input measurement (e.g., coordinate point) contributes to a globally consistent solution (e.g., the physical line that gave rise to that image point). For example, when fitting a set of line segments to a set of discrete image points (e.g., pixel locations), a lack of knowledge about the number of desired line segments may leave the possible solutions for fitting line segments through the image points unconstrained.

[0156] In the context of image analysis, the coordinates (i.e., x, y) of points of an edge segment in an image are known and therefore function as constants in the parametric line equation x cos θ + y sin θ = r, where r and θ are unknown variables. When plotting the possible (r, θ) values ​​defined by each (x, y), the points in Cartesian image space are mapped to curves (i.e., sine waves) in polar coordinate Hough parameter space, as shown by the Hough transform image 1420 in Figure 14C. This point-to-curve transformation is the Hough transform for a line. When viewed in Hough parameter space, points that are collinear in Cartesian image space produce curves that intersect at a common (r, θ) point.

[0157] The Hough transform can be used to identify one or more parameters of a curve that fits a given set of points. In some cases, the Hough transform can also be useful for identifying features (i.e., to detect features with parametric descriptions) and how many of them are present in an image. The curves generated by collinear points in the gradient image intersect at peaks (r,θ) in the Hough transform space. These intersections characterize line segments in the original image. An extraction mechanism can be used to extract local maxima (e.g., intersections) from the accumulator array. For example, one method may include applying thresholding and thinning to clusters of isolated local maxima in the accumulator array image or the Hough transform image 1420. Thresholding in this Hough transform context refers to setting a predefined limit on the maximum value in the accumulator array, which may be greater than or equal to the predefined maximum value.

[0158] In some embodiments, identifying lines in normalized image space 1410 that pass through multiple detected points may include, among other steps, using a Hough transform to map all lines that pass through a given point in a single sinusoidal wave in the Hough transformed image 1420, discretizing Hough space into a number of bins 1424, and incrementing the bin 1424 by one if the sinusoidal wave passes through that bin. Each detected point in normalized image space 1410 is transformed into Hough space, and if the value of a bin 1424 in Hough space exceeds a predetermined threshold, that Hough space point, when de-Houghed, will map to a line that passes through at least that number of points in the image. For example, in a 5×5 grid array of detected points shown in normalized image space 1410, the binning threshold can be set to 4. A point in Hough space represented by a bin 1424 with a binning threshold of 4 will map to a line that passes through at least four points in normalized image space 1410.

[0159] In some embodiments, the data association algorithm may include refining the lines by, for example, rejecting or removing lines that fit poorly to the detected points. Discretization in Hough space may result in lines that may deviate slightly from the optimal line through the set of detected points (as shown in image space 14130 in FIG. 14D ). Alternatively, or additionally, discretization may result in multiple lines that are close to each other. The data association algorithm may include refining the set of optimal lines by rejecting lines that fit poorly. The refinement step may include selecting a line from the set of lines, drawing the line in image space, identifying all points within a predefined distance of the line, performing a simple linear regression to define the line that fits the points best, and removing other lines that are within a certain distance in Hough transform space. These steps may be repeated any number of times as needed.

[0160] Figure 14E represents a Hough transform image 1440 after the line refinement step discussed in connection with Figure 14D above. In the Hough transform image 1440, clusters 1444 correspond to the lines present after the refinement step. One or more cycles of line refinement may be performed iteratively as necessary.

[0161] In some embodiments, associating each identified fiducial in the image with a point in the 2D image of the detected points may include de-Houghing, i.e., mapping points from a polar coordinate image space (e.g., a Hough transform space) to a regular grid (e.g., a Cartesian coordinate image space). This may allow for identification of missing points or points that are not included in the grid. De-Houghing may include identifying groups of parallel lines by identifying groups of all Hough points with similar theta (θ) values ​​and using a clustering algorithm (e.g., k-means clustering) to identify groups of lines with similar theta (θ) values. Figure 14F shows a k-means clustering representation 1450 of the refined lines of Figure 14E in accordance with disclosed embodiments. It should be understood that other clustering algorithms may also be used. In representation 1450, clusters of lines with similar slope (or θ) values ​​may be formed, and the spread of the points in representation 1450 indicates the range of slopes of the lines within the cluster. Thus, representation 1450 shows four groups of substantially parallel lines with clustering based on one dimension (θ). For example, a first group of substantially parallel lines may be represented as having a slope in the range of 0 to 1 radian, a second group of substantially parallel lines may be represented as having a slope in the range of 1 to 2 radians, a third group of substantially parallel lines may be represented as having a slope in the range of 2 to 2.5 radians, and a fourth group of substantially parallel lines may be represented as having a slope in the range of 2.5 to 3 radians.

[0162] From representation 1450, two points from the most populated group and two lines with different θ values ​​can be selected. The intersections of the four lines in image space can be used to generate four points that form a rectangular frame, as shown in representation 1460 of FIG. 14G. The vertices of the rectangular frame may be labeled 0, 1, 2, and 3 (starting from the top-right vertex counterclockwise to the bottom-right corner). In some embodiments, a homography matrix may be calculated that moves points in image space onto an integer grid, preferably a square grid. In the context of this disclosure, a homography is a transformation that occurs between two planes. In other words, it is a mapping between two planar projections of an image using a transformation matrix; multiplying a point in a view by the homography matrix allows for a shift from one view of the same scene to another to find its corresponding position in the other view. In the context of this disclosure, because a homography matrix has eight free variables (each point may include x and y, resulting in a total of eight equations), at least four points may be required to calculate a homography.

[0163] After computing the homography, all points may be mapped to an integer grid using the computed homography. In some embodiments, the mapped points may be rescaled so that their minimum and maximum values ​​lie at the edges of the integer grid 1470 shown in FIG. 14H.

[0164] One way to determine a successful mapping by the association algorithm is to determine whether each point in the image 1460 maps to a distinct, discrete location on the square integer grid 1470. Each reference location on the square integer grid (e.g., the integer grid 1470) may be labeled or numbered with a reference character (e.g., a numeric, alphanumeric, letter, or other suitable character) based on a predefined sequence. The mapping of a point from the image 1460 to the integer grid 1470 may indicate a distance or "offset" between the reference location on the integer grid 1470 and the mapped point. The offset may be expressed in any units or may indicate the actual offset distance between the reference location and the mapped point.

[0165] In some embodiments, associating an identified fiducial in an image with the fiducial's 3D location may further include filtering out false or out-of-range detections. False detections may include detection of a light source that is not recognized as a verified light source, for example, but not limited to, a reflection from an object or a transient light source with similar characteristics. Filtering out false detections may include rejecting points that are more than a predetermined threshold offset distance from the reference location. In some embodiments, the predetermined threshold offset distance may be an absolute integer value, or a fraction, or a percentage of the distance between two adjacent reference locations, or other numerical value.

[0166] Further, in some embodiments, a data association that produces a mapped integer grid (e.g., integer grid with mapping 1470) can be rejected based on the number of identified false positives. For example, if the number of false positives exceeds a predetermined false positive threshold, the data association can be rejected, resulting in no association at all and making the data association algorithm a reliable data association algorithm. Figure 14J shows a detected image 1480 including annotations of light sources detected based on the data association.

[0167] d. Point Tracking Algorithm Reference is now made to FIG. 15A, which illustrates an example image showing shifts between frame points and corresponding new predicted positions using a homography for point tracking, according to disclosed embodiments. Point tracking may be used to generate associations for a new set of detected points. One of several ways to perform point tracking may include propagating existing associations from one image frame to the next based on a previous set of associated points and corresponding pose estimates. In some embodiments, point tracking may include extracting corresponding features between the two sets of frames, computing a homography using the corresponding features, applying the homography to the previously known associations, and mapping the associations to the new detections using a nearest neighbor search.

[0168] In some embodiments, point tracking may involve extracting features or unique characteristics from a previous frame set and identifying corresponding features in a current frame set. In this context, the previous frame set and the current frame set refer to the (n-1)th frame set and the (n)th frame set, where n is an integer. The previous frame set (the (n-1)th frame set) may include a plurality of pixels of an image captured by, for example, a camera mounted on the airborne vehicle when the airborne vehicle is at a position (p-1) at a particular time (t-1), and the current frame set (n0) may include a plurality of pixels of an image captured by the camera when the airborne vehicle is at a position (p0) at a particular time (t0). As used herein, the current frame set refers to the frame set immediately following the previous frame set, such that there are no frame sets between them.

[0169] In some embodiments, two sets of features extracted from the two frame sets can be used to determine a homography matrix configured to transform points in any image to corresponding points in the other image. The determined homography may be configured to shift points in the previous frame set to predicted positions in the next frame set.

[0170] In some embodiments, point tracking may be implemented as a locally associated tracking by applying a determined homography to previously associated detected points in a previous frame set. Alternatively, point tracking may be determined as a pose-based tracking. An exemplary pose-based tracking technique may include using a previously known position estimate, projecting a known fiducial into an image frame using the pose estimate, and applying a homography to the projected points as a prediction of where the points will be in the current frame.

[0171] In some embodiments, as shown in Figure 15B, a nearest neighbor search step may be used to find correspondences between detected points from the detection step and predicted points from the tracking step, allowing for finding associated identifications for each detection. Image 1520 shows an example image generated using a nearest neighbor-based search to identify detections that correspond to known tracked associations, in accordance with disclosed embodiments.

[0172] In the context of this disclosure, temporal filtering refers to separating the frequency components of a time sequence of an image into specific bands or ranges in optical signal processing. Filters used in temporal filtering may be of any standard form, such as finite impulse response or infinite impulse response. Spatial filtering refers to a process that can modify the characteristics of an optical image by selectively removing specific spatial frequencies that constitute the object. In spatial filtering techniques, the Fourier transform of an input function may be operated on by a filter. Spatial filters may be convolution filters (where a kernel is moved across the image) or other forms of spatially oriented filters. For example, a Gaussian blur can be applied to an image to remove high spatial frequencies.

[0173] As previously mentioned, the Active Fiducial Light Pattern Localization (AFLPL) approach, referred to herein as Precision Landing and Takeoff (PLaTO), offers significant advantages for localizing eVTOL aircraft during the approach and landing phases when GPS is difficult, obstructed, or completely unavailable. However, to optimally utilize AFLPL, its potential limitations and approaches to mitigating them must be understood, addressed, and developed. A brief description of the limitations and mitigation strategies is provided herein. i. Geometric Limitations - The primary geometric limitation of AFLPL is the dilution of precision that occurs as the range to the illumination constellation increases. For a fixed focal length camera, the maximum useful horizontal range is approximately 100 times the size of the illumination constellation. A second geometric limitation is that the camera's field of view (FOV) limits the approach trajectory to the landing site. This limitation can be mitigated by using multiple cameras with different field of view and FOV. Because cameras are small, lightweight, and low-power, two or three cameras can be utilized with minimal impact on aircraft cost and performance. ii. Detection and SNR Limitations - In visual and near-infrared systems, background light from the sun and other artificial sources can be significant compared to the illumination intensity of the proposed constellation. AFLPL sources must have sufficient brightness, modulation capabilities, and a distinct wavelength combination to enable detection against a crowded, backlit, but primarily static background. A high percentage of landing scenarios where GPS is unavailable can be achieved by using a combination of visible, near-infrared, and long-wavelength infrared sources and detectors. iii. Data Association—Viewpoint—The n-point algorithm requires that the location of light sources in a constellation be known and that the detection of point sources in the camera's focal plane be associated with each individual light source in the constellation. This data association problem is further complicated by camera motion and looming, which can cause point sources to move out of view. This problem is well-studied and fundamental to many computer vision applications, and many standard algorithms, such as RANSAC, have been developed to solve the association problem. It is important to note that association with fiducials composed of active point sources is significantly less complex than association with visual features from unstructured passive images. The AFLPL approach deals with a small number of distinct point sources with a high probability of detection and favorable placement. Furthermore, association can be facilitated by temporally modulating each fiducial source with a unique code or by spectral matching, where each source emits light at a different wavelength.

[0174] To verify the feasibility of the AFLPL approach, simulation studies and related analyses were conducted to explore the accuracy of localization solutions generated by various viewpoint-n-point (PnP) algorithms under various operational conditions, including the number of light sources in the constellation, the physical size of the constellation, and the image quality of the camera system. Two approaches were used to model fiducial-based localization for eVTOL aircraft. Equations were implemented to model a pinhole camera with a 4k imager (3840 x 2160 resolution) and a 90-degree field-of-view lens. The aircraft trajectory and lighting constellation configuration corresponding to the landing profile shown in Figure 2 were also implemented. These models were used to examine the sensitivity of localization accuracy to variations in the number of light sources in the constellation, the size of the constellation, and the camera's image error, as discussed with reference to Figures 10A, 10B, 11A, 11B, 12A, and 12B.

[0175] Referring to FIGS. 10A and 10B, simulated data plots illustrating the effect of constellation size on localization accuracy in the horizontal and vertical directions, respectively, are shown in accordance with some embodiments of the present disclosure. To study the effect of the number of constellation points on localization accuracy, an eVTOL camera was placed at discrete locations along the landing profile of FIG. 2. At each location, the camera position and altitude were calculated using an embodiment of the proposed system. To generate the plots of FIGS. 10A and 10B, the number of light sources was varied. For each range of light source count, the light sources were evenly distributed within a 40 m x 40 m x 20 m rectangular volume. 1000 randomly placed light source configurations were used to calculate the average localization error. The camera position was compared to the true camera position, and the horizontal and vertical components of the localization error were calculated and averaged over 1000 runs to generate the plots. The positions of the light points within the image frame were rounded to the nearest pixel. As shown in the figures, the localization error decreases as the number of imaged light sources increases. The localization error is less than 10 m at a range of 1200 m and less than 5 m at 500 m when at least 15 light sources are used. As the aircraft approaches the landing target, the localization error drops to the centimeter-level range.

[0176] 11A and 11B show simulated data plots illustrating the effect of light source constellation size on horizontal and vertical localization accuracy, in accordance with some embodiments of the present disclosure. The simulated data plots were generated by varying the horizontal size of the light source constellation while keeping the number of light sources and the vertical size of the constellation constant. The vertical dimension of the constellation was 20 m, and 20 light sources were used. For each constellation size at each camera position, the 20 light sources were evenly distributed within the constellation volume. 1000 randomly placed light configurations were used to calculate the average localization error. The camera position was compared to the true camera position, and the horizontal and vertical components of the localization error were calculated and averaged over 1000 runs to generate the plot. As shown in the figures, the localization error decreases as the constellation size increases. When a horizontal reference line of at least 30 m is used, the localization error is less than 10 m at a range of 1200 m and less than 1 m at 500 m. As the aircraft approaches the landing target, the localization error drops to the centimeter-level range.

[0177] 12A and 12B show simulated data plots illustrating the effect of centroid error on algorithm robustness in the horizontal and vertical directions, according to some embodiments of the present disclosure. When a light point is imaged by a camera, small errors can be introduced by the optics and image sensor. The plots shown in FIGS. 12A and 12B depict the impact of these errors on localization accuracy. The plots were generated by rounding the floating-point pixel locations of the light sources (the true centroid locations on the image plane) to the nearest integer (average error of 0.3 pixels) and then adding various normally distributed errors with averages of 0, 1, 2, or 3 pixels. This was done for each range while keeping the number of light sources (20) and the constellation size (40 × 40 × 20 m) constant. The average localization error was calculated using 1,000 randomly placed light source configurations with added imaging errors. The camera position was compared to the true camera position, and the horizontal and vertical components of the localization error were calculated and averaged over 1,000 runs to generate the plots. As shown in the figure, the localization error increases with the error in the center of gravity position. If a horizontal reference line of at least 30 m is used, the localization error is less than 10 m at a range of 1200 m and less than 1 m at 500 m. As the aircraft approaches the landing target, the localization error drops to the centimeter-level range. The cases with errors of 0.3 pixel and 1.3 pixel meet the horizontal error requirement.

[0178] 12C and 12D, simulated data plots illustrating horizontal and vertical localization errors for a landing trajectory during a simulated approach of an aircraft in accordance with some embodiments of the present disclosure are shown. The algorithm was tested using an eVTOL simulation environment. The localization error increased during the transition from the forward camera to the zenith camera. It is expected that this transition error can be resolved by running the zenith localization algorithm before the transition and fusing its results with the results from the forward camera, rather than abruptly switching between the two.

[0179] Optical Constellation Design

[0180] In some embodiments, data association accuracy may be affected by distortion of the light source pattern due to viewpoint. For example, a light source pattern viewed directly above a landing surface may have a distinct shape compared to a light source pattern viewed from a shallow viewing angle. This discrepancy in the shape of the light source pattern based on viewing angle may be referred to as a “viewpoint transformation.” Many distinguishing features may be lost under viewpoint transformation. Therefore, it may be desirable to design an active light source fiducial pattern and develop an accompanying data association algorithm to identify fiducial pattern points in images captured by a camera mounted on an airborne vehicle. It may further be desirable to create a software pipeline for generating highly accurate pose estimates using bursts of aerial images of a ground-based active light source fiducial pattern. In situations such as dense urban environments or inclement weather, it may be difficult to obtain an estimate of the aircraft's position and orientation. In such situations, performing data association and subsequent pose estimation without prior knowledge or estimate of the aircraft's position and orientation may further exacerbate the problem. Therefore, it would be further desirable to design a fiducial pattern and data association algorithm capable of performing data association and subsequent pose estimation based on isolated snapshot bursts of a fiducial pattern of light sources taken from any viewpoint without additional sensors or measurement mechanisms. The proposed fiducial pattern design and data association algorithm are designed to address one or more of the challenges identified herein. In some embodiments of the present disclosure, the proposed fiducial pattern design and data association algorithm may enable the pose estimation pipeline to continue functioning even in the event of an obstruction or failure that prevents the observation of some constellation light sources in the fiducial pattern or when additional environmental light sources are visible in the camera image.

[0181] Many geometric properties that exist in three-dimensional space are inconsistent when mapped to two-dimensional space under a projective transformation. For example, the length, area, center of gravity, and parallelism in a camera image all depend on the camera's position and orientation relative to the image's subject. However, the intersection ratio remains constant regardless of the viewpoint from which the camera image is taken and serves as a key principle of fiducial constellation design. In some embodiments, fiducial constellation design exploits the projection-invariant property of the intersection ratio. The intersection ratio is a viewpoint-invariant property that can be used for accurate data association. As used herein, "intersection ratio" refers to the ratio of four values, each calculated from a unique subset of features, and is the product of two of these values ​​divided by the product of the other two. For example, the linear intersection ratio is the ratio of the length ratios between collinear points, and the angular intersection ratio is the ratio of the angle ratios between intersecting lines.

[0182] Referring to FIG. 16A, which shows an example line with four points A, B, C, D located at different distances, the linear intersection ratio can be calculated as follows:

number

[0183] Crossing ratios are viewpoint-independent and are constant values ​​for line segments. Different crossing ratios can be obtained depending on the choice of line segments used in the calculation, but they still remain the same regardless of the viewpoint. Crossing ratios are visually invariant. For example, for a set of four points on a line, a total of six crossing ratios can be calculated. These six crossing ratio values ​​can be used to calculate a single invariant specific to the spatial distribution of the four points on the line.

[0184] 16B, which illustrates an exemplary constellation pattern of light sources according to disclosed embodiments. The light sources may be arranged along line segments AA, BB, CC, and DD arranged in a star configuration, such that different angles may be formed by the intersecting lines. In such a configuration, the angle intersection ratio can be calculated as follows:

number

number

[0185] In some embodiments, the landing surface may include a constellation of light sources. A plurality of light sources may be arranged on the landing surface of an airborne vehicle, the arrangement of light sources defining a set of intersecting virtual lines, with light sources positioned on each virtual line and the distance between adjacent light sources on each virtual line being unequal, as shown in FIG. 17A, which illustrates a viewpoint-invariant active light source constellation design 1700 according to an embodiment of the present disclosure. As shown in FIG. 17A, the geometric features of the constellation design 1700 include four lines (AOA, OB, OC, OD) intersecting at a single point O, labeled as lines 1702, 1704, 1706, and 1708. Each of the lines 1704, 1706, and 1708 may be a virtual line connecting five light sources B1-B5, C1-C5, and D1-D5, respectively. Line 1702 may connect ten light sources A1-A10. The different lines in the constellation and their associated light sources can be detected by Random Sample Consensus (RANSAC) or other means. For this set of lines, two unique angle crossing ratios may exist: one calculated using the ordered lines ABCD, and the other calculated using the order BCDA. Since line AOA can be distinguished from the other lines by the number of lights (i.e., 10 for five lights OB, OC, and OD), the correct identity of each other line can be confirmed by calculating the crossing ratio obtained using three lines in clockwise or counterclockwise order from line OA.

[0186] To associate light detected in a camera image with a particular light in a fiducial constellation (e.g., constellation design 1700), a data association algorithm such as grid association, ICP, thin-plate spline, or point tracking, or a combination thereof, may be configured to fit lines to the detected points in the camera image, identify particular lines in the constellation, and identify particular points within each identified line. The data association algorithm may include one or more of the following steps:

[0187] In some embodiments, the data association algorithm may include using RANSAC to determine a best-fit set of four lines with a single intersection point and the correct angle crossing ratio among the points detected in the camera image. Some examples of potential RANSAC sampling techniques, such as random pair sampling and k-nearest neighbor (k-NN) sampling, are shown in Figures 18A and 18B, respectively.

[0188] In some embodiments, using a RANSAC sampling method may include the following steps: (a) sampling four pairs of points from the detected light point cloud and drawing a line through each pair to form four lines. The point pairs can be sampled randomly, or pairs can be drawn from a set of k-nearest neighbors to increase the chances of finding pairs of points that lie on the same constellation line. The type of sampling used for RANSAC (random pairs or k-NN sampling) may depend on how the set of k-nearest neighbors can be efficiently computed; a K-dimensional (KD) tree can be used, whereas point pairs can be randomly sampled. (b) calculating an intersection point represented by a least-squares solution of a system of equations describing the set of four lines sampled in step (a). If the error associated with the solution exceeds a predetermined threshold, i.e., if the point that best fits the set of four lines is far away from each line, then the set of lines does not have an intersection point that is close enough, and the algorithm may return to step (a). Step (c) includes calculating the angular crossing ratio of the set of lines by starting with any line and then including three other lines in clockwise or counterclockwise order after verifying that the set of lines intersects at a single point. As previously mentioned, there may be only two possible angular crossing ratios within the constellation design 1700. Therefore, if the calculated crossing ratio is outside a predetermined error threshold of the expected crossing ratio, the algorithm may return to step (a). In step (d), the number of inlier points for the set of lines is determined by identifying a set of four lines that intersect at a single point and matching it with the expected angular crossing ratio of the constellation, as shown in FIG. 18C. If the number of inlier points exceeds the current best inlier count, the current set of lines may become the best set of lines, and the current inlier count may become the new best inlier count. Step (d) may further include assigning each inlier point to the nearest line in the image.

[0189] In some embodiments, the data association algorithm may further include using the known angular intersection ratios of the constellation design to determine the identity of each line in the set of lines obtained in determining the best fit and correct angular intersection ratio of the four lines. Using the known angular intersection ratios of the constellation design to determine the identity of each line in the set of lines may include performing a line fit for each of the four lines in the best line set using the inlier points assigned to each line in step (d) above. Based on the line fit, the line with the most inlier points may be identified as the reference line (AOA) of the constellation design 1700, and the resulting angular intersection ratio is calculated using lines ordered clockwise or counterclockwise from the constellation reference line. This intersection ratio will match either of the two possible known constellation angular intersection ratios, and therefore the set of lines used to calculate this intersection ratio should be identified as the lines used for the matched pre-calculated intersection ratio, as shown in FIG. 18D.

[0190] In some embodiments, as shown in FIG. 19 , the data association algorithm may further include voting for each point in line 1900 for each line in the constellation using the known linear crossover ratio of the constellation design to determine the identity of each point. Line 1900 may connect at least four light sources. In other words, line 1900 may include at least four points. In some embodiments, line 1900 may include five or more, six or more, seven or more, eight or more, nine or more, ten or more, or any suitable number of points. As shown in FIG. 19 , line 1900 connects six points (points 1-6). For each line, the points may be sorted by distance from a constellation vertex or intersection point (e.g., point 1 on line 1900). For every subset of four points on line 1900, the linear crossover ratio may be calculated and compared to the pre-calculated known linear crossover ratio of the constellation line. 19 shows a table 1910 of known crossover ratios and a table 1920 of calculated crossover ratios for subsets of four points. The table 1910 of known crossover ratios may include the crossover ratios for the subsets of four points of each line 1702, 1704, 1706, or 1708. The table 1920 of calculated crossover ratios may include the crossover ratios for the subsets of four points of line 1900.

[0191] When comparing the known crossover ratios and the calculated crossover ratios, a voting table 1930 may be generated. The voting table is a grid of rows and columns of vote numbers. If the calculated linear crossover ratio (e.g., the crossover ratio of line 1900) is within a predetermined threshold of any of the known crossover ratios of lines 1702, 1704, 1706, or 1708, each point used in the crossover ratio calculation receives a vote over the corresponding point in the set used for the known crossover ratio.

[0192] In some embodiments, a voting strength for each fiducial point may be calculated. As used herein, "voting strength" refers to the ratio of the number of votes for the candidate point with the most votes to the number of votes for the candidate point with the second most votes. If the voting strength of a fiducial point exceeds a predetermined voting strength threshold, the fiducial point may be assigned the identity of the candidate point that received the most votes. As an example, the known intersection ratio of line B1B2B3B4 is 1.35, which is closest to the calculated intersection ratio of 1.36 for line 1900 connecting points 1, 3, 4, and 5 (labeled line 1345 in table 1920). If the difference between the known intersection ratio and the calculated intersection ratio (1.35-1.36=-0.1) is within a predetermined threshold difference, each point used to calculate the intersection ratio (e.g., points 1, 3, 4, and 5) may receive a vote for the corresponding point in the set used for the known intersection ratio. Corresponding points refer to the location of the point on the line relative to the vertex. Point B1 on line 1704 corresponds to point 1 on line 1900, point B2 on line 1704 corresponds to point 2 on line 1900, point B3 on line 1704 corresponds to point 3 on line 1900, point B4 on line 1704 corresponds to point 4 on line 1900, and point B5 on line 1704 corresponds to point 5 on line 1900.

[0193] Referring to FIG. 17B , an exemplary constellation design 1750 of light sources is shown, in which the same number of light sources (e.g., light sources 1751, 1753, 1755) may be arranged on each of imaginary lines 1752, 1754, 1756, 1758, 1760 that intersect at point 1780. Multiple light sources may be arranged on the landing surface of an airborne vehicle, with the arrangement of light sources defining a set of intersecting imaginary lines, with light sources arranged on each imaginary line and the distance between adjacent light sources on each imaginary line being unequal, as shown in FIG. 17A , which illustrates a viewpoint-invariant active light source constellation design 1750 in accordance with an embodiment of the present disclosure. While only three light sources are labeled on line 1752, it should be understood that there may be more light sources (shown as small, unlabeled circles in FIG. 17B ) on each line. The linear intersection ratio of each imaginary line may not depend on the angle of viewpoint. While each imaginary line is shown to include an equal number of light sources, it should be understood that other combinations and permutations are possible. For example, five light sources may be arranged on one imaginary line, and six light sources may be arranged on an adjacent imaginary line.

[0194] Lines 1752, 1754, 1756, 1758, and 1760 may be virtual lines connecting light sources. For example, virtual line 1752 may connect at least light sources 1751, 1753, and 1755. All virtual lines may intersect at virtual intersection point 1780. In this case, the intersection ratio of the angles formed by the intersecting lines is invariant across projections. Different lines in the constellation and their associated light sources can be detected by RANSAC or other means. Intersection points can be determined by the intersection of multiple lines. The intersection ratios of (A, B, C, D) and (B, C, D, E) can be used to determine the order of the angle distribution of the lines. The order of points along individual lines can also be calculated using linear intersection ratios, as previously described in connection with FIG. 17A.

[0195] In some embodiments, each active light source (e.g., light source 350 in FIG. 3) may be configured to self-identify via a modulation scheme. For example, the light source may be modulated to communicate data. The data, in some embodiments, may include an encoded authentication or identification signal that, when received by the photodetector or camera 311, may be used to locate the active light source or identify the landing surface.

[0196] Rapidly deployable optical constellations In some embodiments, the landing surface may include a portable landing surface. The portable vertiport shown in FIGS. 43A and 43B may include a rapidly deployable landing surface 4310 including a redeployable landing mat, cloth, or tarp, or multiple light sources 4325. In some embodiments, the rapidly deployable landing surface 4310 may include a constellation of IR light sources. In some embodiments, the rapidly deployable landing surface 4310 may include a combination of point and linear light sources (as described with reference to FIGS. 30 and 31). This may be particularly useful when landing in uncooperative locations where limited or temporary landing infrastructure may be required, such as for military operations, firefighting, disaster relief, or medical support operations. In some embodiments, the portable vertiport may include an integrated, battery-powered active light source that can be remotely activated, deactivated, or modulated.

[0197] In some embodiments, the rapidly deployable landing surface 4310 can include a constellation of light sources arranged on a rollable mat, which can be carried in a backpack, for example. The rapidly deployable landing surface 4310 may be deployed to ad-hoc landing sites for emergencies, adverse situations, rescue operations, etc. In FIG. 43B, the rapidly deployable landing surface 4340 can be a flexible net or mesh with light sources 4325 woven or clipped onto it.

[0198] In some embodiments, estimating the pose of the aircraft based on images captured from a camera mounted on the aircraft may include communicating a constellation configuration of light sources 4325 arranged or located on the rapidly deployable landing surface 4310 or 4340 to an onboard processor associated with the aircraft. The constellation configuration may be determined by calibrating the relative positions of the light sources using ultra-wideband (UWB) signals for auto-ranging between the light sources. As used herein, ultra-wideband signals may be used to transmit information over a wide bandwidth (>500 MHz). This allows for the transmission of large amounts of signal energy without interfering with conventional narrowband and carrier wave transmissions within the same frequency band.

[0199] In some embodiments, the light sources 4325 may be woven into a rapidly deployable landing surface 4340 (e.g., netting or flexible mesh), which can be spread over an ad-hoc landing site. The landing site may be uneven, bumpy, or non-coplanar, and the rapidly deployable landing surface 4310 or 4340 may adapt to the landing site. The placement of the light sources 4325 may be at known locations on the rapidly deployable landing surface 4340, and the general configuration or location of the constellation of lights may be known accordingly. In some embodiments, the positions of the lights are calibrated using, for example, multiple aerial images acquired by an aircraft. Alternatively, or additionally, the configuration of the constellation may be learned using UWB signals for auto-ranging between the lights and calibrating the relative positions of the lights.

[0200] Automatic generation of optical constellation patterns In some embodiments, a constellation pattern of light may be automatically generated to maximize one or more properties of the constellation. For example, a predefined metric may be used to design a constellation to maximize the ability to distinguish light for detection and data association purposes, effectively maximizing the accuracy and robustness of the data association process. One possible metric is the variance of the linear or angular crossing ratios between multiple lines.

[0201] Incorporating uncertainty into optical position estimation As previously mentioned, the data association step involves associating the location of the light source in the image plane with the physical location of the light on the ground. In some cases, associating the light source between the image and the ground can be challenging because there can be uncertainty in the physical light location due to uncertainty in the light image location (on the image plane) due to measurement error or image error. In some embodiments, incorporating uncertainty information from these location measurements into the data association algorithm allows for the probability of a correct match to be calculated, which may enable the data association algorithm to utilize a confidence metric to facilitate the decision-making process.

[0202] Pose recovery / pose estimation In some embodiments, determining the position and orientation of the airborne vehicle may include detecting at least one light source in the image, determining which of the at least one light source arranged in a predetermined pattern the detected light source is, and determining the position and orientation of the airborne vehicle based on the determination of which of the at least one light source arranged in the predetermined pattern the detected light source is. In a preferred embodiment, the localization algorithm (e.g., PnP) requires that at least four light sources are detected in the image plane (and their pixel locations are determined) and that the pixel locations of those light sources are properly associated with physical light sources arranged in a predetermined pattern (and known locations) on the ground, so that an estimate of the pose of the camera and / or the airborne vehicle (e.g., camera 311, airborne vehicle 310) may be generated.

[0203] In a further preferred embodiment, the localization algorithm may require that at least five light sources be detected in the image plane and that their pixel locations be correctly associated with physical light sources arranged in a predetermined pattern. Generally, as previously discussed in connection with Figures 10A and 10B, the accuracy of camera and / or aerial vehicle pose estimation improves as the number of light sources detected in the image plane and correctly associated with corresponding physical light source locations in the predetermined pattern increases.

[0204] In some embodiments, determining the position and orientation of an airborne vehicle may include detecting three light sources in an image plane and correctly correlating them with corresponding physical light source locations in a predetermined pattern. Standard triangulation techniques can be used to identify two potential locations at the three points. In the case of a landing surface or vertiport, one of the two potential solutions may be eliminated because it is below the landing surface, thereby narrowing the solution to a single location.

[0205] In some embodiments, determining the location of the airborne vehicle may include detecting two light sources in an image plane and correctly associating them with corresponding physical light source locations in a predetermined pattern when orientation information such as yaw, pitch, and roll, and a 3D gravity vector are known. In some embodiments, determining the location and / or orientation of the airborne vehicle may include detecting a single light source in an image plane and correctly associating them with corresponding physical light source locations in a predetermined pattern when aircraft altitude information is known.

[0206] In some embodiments, one or more light sources in a constellation design (e.g., design 1700 or 175) may not be observable and accurately identifiable due to adverse conditions, including, but not limited to, severe weather, dense urban environments, poor light penetration, additional ambient light, etc. In such scenarios, it may be desirable to continue to perform data association and pose estimation for an airborne vehicle approaching or taking off from a landing surface. The proposed pose estimation algorithm and system addresses some of the challenges described above.

[0207] Referring to FIG. 20A, a process flowchart illustrates an exemplary method 2000 for estimating the pose of an airborne vehicle in accordance with an embodiment of the present disclosure. Method 200 may be performed using a pose estimation algorithm in a precision landing and takeoff system and data communication system (e.g., PLaTO system 300 of FIG. 3). For example, a processor (e.g., processor 312 of FIG. 3) may be configured to execute the pose estimation algorithm and programmed to implement the steps of the pose estimation algorithm. It will be understood that the steps performed in method 2000 may be reordered, added, deleted, or edited as appropriate. The computer-implemented pose estimation algorithm may cause system 300 (including a camera, photodetector, microprocessor, memory, or storage device onboard the aircraft) to perform the following steps of pose estimation method 2000:

[0208] In step 2010, a camera (e.g., camera 311 of system 300) is configured to capture a continuous stream of images. A camera application programming interface (API) may be configured to receive the continuous stream of camera images at a frame rate of 100 frames per second (fps). The camera API may be further configured to send three consecutive camera image frames to the photodetector. In some embodiments, the fiducial lights in the constellation design may flash at a frequency such that each constellation light is on for at least one of the three frames and off for at least one frame.

[0209] In step 2020, the light detector is configured to generate an output including pixel locations of detected light sources based on the frames received from the camera in step 2010. Generating the output includes constructing maximum and minimum images consisting of the respective maximum and minimum grayscale intensities at each pixel across the three images and subtracting the minimum image from the maximum image to remove ambient background light. The flashing of light causes the constellation to appear in high contrast against the background and be easily detected in the minimum-maximum image. Subtracting the minimum image from the maximum image includes subtracting the intensity of each pixel in the minimum image from the intensity of each corresponding pixel in the maximum image. In some embodiments, flashing may include adjusting the intensity of the light between a "minimum" intensity and a "maximum" intensity. The minimum intensity can include zero (light off) or any intensity lower than the maximum intensity of the light source, so that the difference between the maximum and minimum is discernible by the detector.

[0210] In step 2030, a data association algorithm is used to identify which constellation of light corresponds to the detected light source based on the received pixel location of the detected light in the min-max image from the photodetector. The data association algorithm performed in this step may be the algorithm previously described with respect to Figures 18A-18D and 19, or any other suitable data association algorithm. In some embodiments, determining a best-fit set of four lines with a single intersection point, correct angle intersection ratio, among the points detected in the camera image using a RANSAC sampling method may be performed in parallel to increase efficiency and reduce the time required for RANSAC to determine a set of lines with high confidence.

[0211] In step 2040, a pose estimation API is configured to receive pixel locations of the identified constellation points in the camera image and to perform an iterative viewpoint-n-point (PnP) algorithm to generate a pose estimate for the camera in the constellation coordinate system.

[0212] The pose estimation pipeline is configured to run end-to-end in less than 30 milliseconds (i.e., at a frequency of 33 Hz or greater), allowing it to generate real-time pose measurements when acquiring camera images at 100 fps. During real-time, the camera API may be configured to run separately from the data association and pose estimation API, resulting in a more efficient pipeline as the light detection, data association, and pose estimation algorithms run on the current set of images while acquiring the next burst of images.

[0213] Example - Simulation and Hardware Testing of Fiducial Constellation, Data Association, and Pose Estimation Pipelines The fiducial constellation, data association algorithm, and pose estimation pipeline described above were tested both in simulation and in hardware. Hardware results were obtained by running the pose estimation pipeline on an Intel NUC mounted on a hexacopter using a camera equipped with an infrared (IR) filter to image an IR light source in the form of a fiducial constellation placed on the ground. To test the robustness of the data association algorithm and pose estimation pipeline to camera viewpoints, several trajectories were conducted, flying around the constellation at distances of up to 200 meters. The position estimates calculated by the pose pipeline were compared with true measurements provided by a real-time kinematic (RTK) GPS sensor mounted on the hexacopter, and the results of the tested trajectories are shown in Figure 20B. As shown in Figures 21, 22, and 23, the accuracy of the pose pipeline was within a few percent of the camera's distance from the constellation, which was also typical for the other tested trajectories.

[0214] Figure 21 shows altitude estimate 2100A and corresponding error 2100B plotted with RTKGPS truth data for the flight trajectory shown in Figure 20B. Figure 22 shows north estimate plot 2200A and corresponding error plot 2200B as a function of ground truth for the flight trajectory shown in Figure 20B. Figure 23 shows east estimate plot 2300A and corresponding error plot 2300B as a function of ground truth for the flight trajectory shown in Figure 20B.

[0215] Referring to FIG. 24A, an example of a random constellation pattern of active light sources on the ground is shown, in accordance with some disclosed embodiments. In some embodiments, the constellation design 2400 may include randomly placed light sources on the ground. The randomly placed light sources may be IR light sources. In the context of autonomous landing, such as landing of an eVTOL airborne vehicle, poor visibility due to dense urban environments or poor weather or low light conditions can pose significant challenges to safe landing and takeoff.

[0216] The random point constellation design 2400 shown in FIG. 24A may be based on fiducial markers called random dot markers (RDMs), which implement a Locally Likely Range Hashing (LLAH) algorithm to identify randomly placed points. The LLAH algorithm is naturally robust to occlusions due to the clustering properties of the method used to calculate the descriptors. In the context of eVTOL landing and takeoff in dense environments, occlusions may occur due to dead light, obscured light, or light that temporarily falls out of frame as the aircraft approaches the landing surface. Furthermore, constellation placement on a landing pad may be feasible because the points are randomly placed and are not restricted to square or rectangular tags or any particular shape.

[0217] In some embodiments, the LLAH random point identification method may include the steps of constellation design, keypoint registration, and keypoint acquisition. Designing the constellation may include creating a random dot marker by generating random x and y coordinates for N points that fit within a selected marker size. In some embodiments, newly generated points that overlap with existing points may be rejected or excluded from consideration. Due to the sensitivity of the crossover ratio equation to point location, a random distribution of points naturally results in a unique crossover ratio. While other fiducial markers may be constrained by shape, random dot markers can take any shape as long as the dots are placed on a plane.

[0218] In some embodiments, keypoint registration in the LLAH algorithm can benefit from a cross ratio to ensure invariance to viewpoint transformations. While affine invariance may be used due to the small number of feature points used to compute the descriptor, using a cross ratio may offer some advantages over affine invariance. For example, in this context, a low approach angle to the landing pad may result in severe viewpoint distortion of points, and it may be desirable not to assume an affine transformation for local clusters of points. Furthermore, because the number of fiducial points (e.g., infrared light sources on the ground) may be relatively small, affine invariance may be redundant or even inefficient in some cases.

[0219] In some embodiments, a descriptor may be calculated for each keypoint in the constellation to uniquely identify the point using the LLAH algorithm. A descriptor, as used herein, is a sequence of discretized cross ratios. Figure 24B shows an example constellation including keypoint p and n nearest neighbors. Computing the descriptor may include identifying n nearest neighbors for each keypoint p in the constellation and selecting m point combinations from the n nearest neighbors as shown in Figure 24B. From the selected m points, a cross ratio may be calculated using a combination of five points. As implemented herein, keypoint p may be one of the five points used to calculate the cross ratio.

[0220] FIG. 24C illustrates creating a discretized crossover ratio sequence using a combination of five points from m=7 points to calculate the crossover ratios. In some embodiments, the crossover ratios may be discretized, and the discretized crossover ratio sequence from each combination may be used as a descriptor for keypoint p. In this method, the discretization boundaries may be chosen by calculating all crossover ratio combinations for the constellation, sorting them, and dividing them into a selected number of buckets. The upper and lower limits of a bucket may be defined by the crossover ratios with the lowest and highest values ​​in that bucket. This method of discretizing the crossover ratios is used because, for a single constellation, the crossover ratio values ​​are not uniformly distributed. Typically, there are many crossover ratios with low values ​​(less than 20) and very few crossover ratios with high values ​​(more than 100). As a result, a finer range may be desired when discretizing the low values.

[0221] Referring to FIG. 24D, discretization of cross ratios to create a discretized cross ratio sequence is illustrated in accordance with some disclosed embodiments. In some embodiments, the discretized cross ratio sequence is used as a descriptor instead of the cross ratio value. As shown in FIG. 24D, two cross ratio sequences (e.g., Set 1 and Set 2) may be generated. When the cross ratios are discretized, Set 1 and Set 2 have the same cross ratio values. However, the sequence in which the values ​​appear may be unique. For example, even if the discretized cross ratios in Set 1 are (0, 2, 0, 3) and Set 2 are (2, 3, 0, 0), the sequences of cross ratios within the sets are different. To facilitate reproducing the sequence during acquisition, the nearest neighbors of each keypoint are sorted in clockwise order before any combination or descriptor is calculated. Each descriptor has a dimension of mC5, and each point has nCm descriptors. As shown in FIG. 24E, a hash index may be calculated from each descriptor, and the keypoint ID and marker ID may be stored in the hash index along with the descriptor.

[0222] In some embodiments, the LLAH algorithm may further include keypoint acquisition. To perform marker acquisition using matching, a descriptor can be calculated for each detected point in the image using the same method described previously. However, because the orientation of the constellation in the camera view may be different from that during the registration step, simply sorting the points clockwise before calculating the sequence may not yield accurate results. The first point in the sorted nearest neighbors of the stored constellation may be different from that of the live constellation. As a result, all n clockwise orderings are calculated and used for voting. For example, if the original clockwise ordering of points was a, b, c, d, e, f, the live acquisition must also calculate descriptors such as b, c, d, e, f, a.

[0223] The descriptor may be used to calculate the hash index needed to look up the table. A vote can be taken for each keypoint ID and candidate marker ID found in the table. For each candidate marker, a homography is calculated using RANSAC to confirm a match. In this case, there is only one candidate marker, so once a vote has been taken for each keypoint, or a number of keypoints above a certain threshold have been confirmed, the homography can be calculated. At this point, we have identified the keypoints.

[0224] In some embodiments, all points (e.g., active light sources) may be non-coplanar or non-collinear. In such cases, the area cross ratio (ACR) may be used. For example, using any five points, the ACR can be calculated from the ratio of the areas of the triangles defined by three of the five points as follows:

number

[0225] As noted above, an example value for the intersection ratio can be determined using the equation: Similar to the linear intersection ratio, for a particular constellation of five non-collinear, non-coplanar points, six unique area intersection ratios can be determined. A single invariant value can be calculated from the six area intersection ratios.

[0226] In some embodiments, an area intersection ratio algorithm may be used for data association, which may include the following steps: 1. Given a set of points to associate, select five unassociated or arbitrary points. 2. Compute the single invariant value described above. 3. Find the closest invariant value to the calculated value from a pre-calculated table. 4. Follow the associated entry into a second pre-computed table specific to that single immutable value. 5. Using the cross ratio values ​​calculated in step 2, find the set of values ​​that most closely match to determine the order of the points. 6. Based on the order of points used in steps 1 and 5, the identities of all five points can be determined. 7. Remove those five points from consideration and repeat step 1. 8. If fewer than five points exist, previously identified points can be used to complete the set of five points.

[0227] In some embodiments, in a table matching technique for data association, a pre-computed table may be formed by enumerating all possible sets of five points in a 3D constellation with known geometry. The resulting j-invariant region crossing ratio may be calculated when each of the five points in each set is selected as a center point. These values ​​form a table whose rows correspond to the IDs of the center points and whose columns correspond to sets of four non-center points in the five groups. It should be understood that some cells in the table are empty because an individual point cannot be both a center and a non-center point in a five-group. One method for completing the data association process may include searching this table and matching j-invariant values ​​calculated from observed 2D image points to pre-computed values ​​in the table. By utilizing already matched points in a table traversal algorithm, the possible j-invariant values ​​in the table that are potential matches at a particular time step can be reduced. This allows for efficient region growing, such as first finding and identifying a single set of five points, and then incrementally increasing the set under consideration by one point at a time to find the best match. Additionally, uncertainty propagation techniques may be used to account for pixel and constellation calibration uncertainties when determining the threshold for matching the j-invariant value.

[0228] In some embodiments, data association may be performed using the Hungarian Association Matrix or the Munkles algorithm. In the Hungarian Association Matrix technique, an example 5x5 matrix of scores may be used, as shown in Table 1 below. [Table 1] Table 1. 5x5 matrix of scores

[0229] They can be converted to costs by subtracting a large number, say 1.0, from everything, and the results can be obtained as shown in Table 2 below. [Table 2] Table 2.

[0230] In the next step, we can subtract the minimum cost from all rows as shown in Table 3 below. [Table 3] Table 3.

[0231] In the next step, we can subtract the minimum cost from all columns as shown in Table 4 below. [Table 4] Table 4.

[0232] In the next step, we can draw as few row or column lines as possible to connect all the zero values ​​in Table 4. From the non-zero elements of the matrix shown in Table 5 below, we find the minimum value. [Table 5] Table 5.

[0233] The next step is to subtract the minimum value (e.g. 15 in Table 5) from all immovable elements and leave the rest as they are. The resulting table is shown in Table 6 below. [Table 6] Table 6.

[0234] The next step is to identify the points in the table. If any points are missing, repeat the steps that generated Table 5. [Table 7] Table 7.

[0235] In a pose recovery or pose estimation algorithm, the physical pose of the airborne vehicle may be estimated based on points associated with those identified in the data association step above. In some embodiments, the relative pose (attitude and position) of the camera with respect to the landing surface may be calculated using a viewpoint-n-point (PnP) algorithm. Various implementations of the PnP algorithm are available (e.g., in the OpenCV library). The accuracy of the PnP calculation may be improved when the constellation points are not coplanar. In some cases, the vertiport constellations may be coplanar, allowing for exploration of the robustness of various PnP approaches to coplanarity. The PnP calculation is relatively fast compared to the detection and association steps and does not limit the speed of the pose recovery pipeline (detection, association, pose recovery). PnP generates a camera pose estimate at the camera frame rate. Since the camera pose relative to the aircraft is known, the aircraft pose relative to the landing site can be calculated from the camera pose information.

[0236] PnP techniques work by optimizing the 3D pose (position and orientation) of the camera while minimizing the reprojection error of 2D points observed in the image to the 3D points reprojected onto the image. Various algorithms exist to solve this optimization, which may be solved by nonlinear least-squares techniques.

[0237] In addition to geometry-based approaches, it may be desirable to fuse PnP-type pose solutions with IMU information using approaches such as an extended Kalman filter (EKF). This fusion approach may offer several advantages, including allowing pose solutions to be calculated at update rates higher than those of the onboard cameras, providing increased redundancy and robustness to the pose recovery process, and allowing outlier PnP solutions to be rejected in a mathematically rigorous manner.

[0238] In some embodiments, a tightly coupled moving horizon estimation (MHE) formulation may be used in which raw camera images and inertial measurement unit (IMU) information are processed simultaneously. The MHE formulation approach is less sensitive to nonlinearities and can be more accurate than a loosely coupled EKF approach, while retaining the same advantages.

[0239] In some embodiments, other image-based navigation aids are used to improve the accuracy of the system, such as using visual odometry, optical flow, or intermediate homography information derived from camera-generated images. Measurements of the aircraft's speed and attitude rate of change can be fused with pose estimation (i.e., using Kalman filters or other sensor fusion techniques) to improve the overall accuracy of the system.

[0240] In some embodiments, one or more characteristics of the light emitted from the active light source can be used to improve detection by background subtraction. Referring now to Figures 26A-26C, exemplary waveforms representing intensity modulation and camera shutter speeds used in encoding / decoding information algorithms are shown in accordance with some embodiments of the present disclosure.

[0241] For background subtraction, it may be desirable to have one frame with the light on at 100% intensity and another frame with the light completely off or at 0% intensity. The maximum difference in the intensity of the light emitted by the light source allows for subtracting each pixel between the two frames and removing all steady light sources. One way to do this is to turn the light on and off at half the camera's shutter speed, as shown in Figure 26A. In Figure 26A, Tm denotes the modulation period and Ts denotes the shutter period. However, such techniques may be limited by timing consistency issues. For example, the camera's sampling rate (also referred to herein as the camera capture rate) may coincide with the transition of the active light source, as shown in Figure 26B. In such a case, because the camera exposure is not instantaneous, the camera may capture half on and half off, resulting in a pixel value of 50%. This not only reduces the total signal intensity, but also impacts the background subtraction algorithm, as all frames will have the same 50% value, potentially rendering the fiducial undetectable.

[0242] To mitigate issues related to timing alignment between the camera capture rate and the active light source transitions, the camera capture rate can be synchronized with the modulation frequency of the light source. However, the synchronization process presents several challenges, notably the need to synchronize all lights together so the cameras can synchronize, and synchronizing to the pulses can be complicated because a fiducial must be identified before synchronization can occur. While feasible, this can add significant complexity to the algorithm.

[0243] In some embodiments, as shown in Figure 26C, the timing mismatch between the camera capture rate and the active light source transition can be overcome by flashing at a rate different from the shutter speed. This solves the problem of 50% pixel values ​​in each frame, but does not completely eliminate partial frames, potentially introducing two consecutive frames with the same exposure rate (shown as two consecutive dark bars in Figure 26C). One way to mitigate the same exposure problem is to increase the sampling size from two to three images. In that case, for background subtraction, rather than subtracting frame 2 from frame 1, take the maximum and minimum pixel values ​​at a particular pixel location across all three images and subtract them. In a given set of three images, one image could be at 100% intensity, another at 0% intensity, and the third frame could have a partial exposure or two consecutive frames with the same exposure rate. By setting a flash frequency that is not a multiple of the shutter speed, it can be mathematically shown that each batch of three images will have at least one image on and one image off.

[0244] In some embodiments, the camera capture rate is at least 100 frames per second. In some embodiments, adjusting the camera capture rate is based on modulation of the light source. Adjusting the capture rate may include synchronizing the camera capture rate with the modulation rate of the light source. In some embodiments, the blinking rate of the light source is about 30 Hz. In some embodiments, the blinking rate of the light source may be adjusted based on the camera capture rate.

[0245] In some embodiments, the controller is configured to adjust the capture rate of the camera based on the modulation of the light source. The controller may be further configured to adjust the blinking rate of the light source based on the capture rate of the camera. The controller may be further configured to adjust the bit rate of the camera, wherein the bit rate of the camera is 10 Hz or greater.

[0246] In some embodiments, the controller is further configured to generate an output signal based on the output signal from the camera, the output signal including information related to the position and orientation of the airborne vehicle. The information related to the position of the airborne vehicle may include GPS coordinates of the airborne vehicle. The controller may be further configured to transmit the information related to the position and orientation of the airborne vehicle to an external processor.

[0247] In some embodiments, one or more characteristics of the light emitted from the active light source can be used to transmit information related to the light source and / or landing surface information related to the light source. Conventional techniques for data transmission have several drawbacks, including, but not limited to, that a universal clock signal may be required to trigger sampling, that binary values ​​of on=1 and off=0 may be insufficient, and that clockless transmission patterns require synchronous sampling, among others.

[0248] In some embodiments, the duty cycle of an active light source can be adjusted. An example of a bit-wise transmission pattern is shown in which a 1 indicates a 70% duty cycle and a 0 indicates a 30% duty cycle. However, the fiducial alternates duty cycle at a rate slower than the light source's blinking rate. For example, the binary value 9 is represented as four bits 1001, as shown in FIG. 27A. Once the fiducial is located by the camera, the light intensity can be averaged over multiple samples to calculate the average duty cycle for that period, as well as the bit value assigned to that period, as shown in FIG. 27B. The transmission and averaging can be tailored to suit transmission rate and noise robustness requirements. Alternatively, the same averaging technique can be used to calculate the average duty cycle, with 1s or 0s represented by rising or falling edges, as shown in FIG. 27C.

[0249] In some embodiments, the camera capture rate and the active light source blinking frequency can be synchronized, as shown in FIG. 26B, to reduce errors that occur when images are captured during transitions. In some embodiments, each active light source can be synchronized by a synchronization pulse transmitted over a connected wired network. In some embodiments, the synchronization pulse can be transmitted wirelessly through RF transmission. The frequency of the synchronization pulse can be appropriately selected. In some embodiments, synchronization can be achieved via an optical sensor associated with each active light source. In such cases, each active light source must be able to detect optical transmissions from at least one other light source. Once each light is synchronized with its neighboring light sources, all active light sources can be synchronized.

[0250] In some embodiments, when the active light source is synchronized, the camera can be synchronized with the active light source by a synchronization pulse transmitted wirelessly via RF transmission. In some embodiments, an on-board processor (e.g., processor 312) may adjust the camera frame rate based on the quality of the reference detection.

[0251] Associative Synthesis As previously mentioned, multiple algorithms are used for data association to associate each identified fiducial point in a 2D image with the 3D location of a fiducial point on the ground. These algorithms include ICP, TPS-RPM, point tracking, linear intersection ratio, angular intersection ratio, and grid association. While each data association algorithm offers some advantages when used alone, it can also pose challenges. Due to their different strengths and applicability, it may be desirable to combine two or more data association algorithms to generate more reliable and robust associations for use in precision landing and takeoff of eVTOLs in GPS-unavailable environments.

[0252] As shown in Figure 28, a flowchart illustrates an exemplary method 2800 for data association and synthesis according to a disclosed embodiment. The exemplary method illustrated in Figure 28 includes the steps of capturing images of vertiport light sources (fiducials) placed on the ground using a camera mounted on an aircraft, identifying a pixel location of each fiducial in the image (detection step), associating each identified fiducial in the two-dimensional image with the fiducial's three-dimensional location on the ground (data association step), determining the aircraft's orientation and position (pose estimation step) and validating the pose based on the association, performing one or more sampling algorithms (e.g., RANSAC) to filter out outliers, and generating the aircraft's pose by performing a viewpoint-to-n-point transformation and applying a pose filter.

[0253] In some embodiments, the data association synthesis pipeline may include executing one or more data association algorithms to generate associations between fiducials identified in the 2D imagery and the 3D locations of the fiducials on the ground. In preferred embodiments, two or more data association algorithms may be executed to generate the associations. The generated associations may be compiled, for example, in a data storage server or memory, to form an aggregated data list or larger information set related to the generated associations.

[0254] In some embodiments, for each data association algorithm executed, a PnP algorithm may also be executed to accurately determine the pose (attitude and heading) generated based on the association. Additionally, or alternatively, for each data association algorithm executed, the determined pose may be verified by, for example, verifying that the aircraft's position is above the ground, verifying that the aircraft is within a certain distance from a designated vertiport or landing surface, verifying that the aircraft is pointing in the correct general direction, among other things. After the determined pose is verified, the associated information may be added to an aggregated data list.

[0255] The data association and synthesis pipeline may further include identifying all unique points from the aggregated data list to form a second aggregated data list and performing a sampling technique (e.g., RANSAC sampling) or similar algorithm to remove outliers. The third aggregated data list containing the relevant information after removing the outliers may form a final aggregated data list and be used to perform a full PnP transformation to generate a pose estimate for the aircraft. In some embodiments, one or more pose filters may be used to generate the pose estimate for the aircraft, including, but not limited to, a Kalman filter, an extended Kalman filter, or other suitable pose filters.

[0256] As shown in Figure 29, a flowchart illustrates an exemplary method 2900 for data association and synthesis according to a disclosed embodiment. The exemplary method illustrated in Figure 28 includes the steps of capturing images of vertiport light sources (fiducials) placed on the ground using a camera mounted on an aircraft, identifying a pixel location of each fiducial in the image (detection step), associating each identified fiducial in the 2D image with the fiducial's 3D location on the ground (data association step), determining the aircraft's orientation and position (pose estimation step) and validating the pose based on the association, performing one or more sampling algorithms (e.g., RANSAC) to filter out outliers, and generating the aircraft's pose by performing a viewpoint-to-n-point transformation and applying a pose filter.

[0257] Compared to method 2800, in some embodiments, each data association algorithm that is run may be treated as an independent sensor, as shown in Figure 29. For each data association algorithm that is run, a full PnP transformation and RANSAC sampling may be performed before combining all measurements using a Pose filter (e.g., a Kalman filter).

[0258] Data association algorithms such as grid association and TPS-RPM may be computationally intensive and iterative in nature, respectively. To overcome these and other issues, an association pipeline may be implemented with dual modes: a "lost in space" mode and a "tracking" mode.

[0259] The lost-in-space mode may be useful when prior information is unavailable, such as during the aircraft's maiden flight or when attitude information is lost due to a newly installed light source, loss of connectivity, or transmission challenges. The lost-in-space mode may operate at a slow rate and may execute grid association and TPS-RPM algorithms to associate detected light with known patterns, thereby providing a set of associated points to a pose-n-pose (PnP) algorithm configured to generate a pose estimate. Based on the confidence level of the association provided by the "lost-in-space" mode, the processor may be configured to switch to a tracking mode that operates at a higher rate than the lost-in-space mode.

[0260] Tracking mode may be obtained when prior information is available or from lost-in-space mode. In tracking mode, tracking algorithms, including but not limited to local related point tracking, pose-based point tracking, ICP, etc., may be initialized with previously calculated poses and associations. These algorithms may generate associations that can be fed into pose estimates individually or as a combined superset. Such pose estimates can be output by the system at a higher rate.

[0261] To further improve the correlation throughput, after switching to tracking mode, a lost-in-space correlation algorithm can be run in parallel to provide corrections. Because the correlation algorithm does not depend on previous states and does not accumulate errors, the overall pose estimation may be improved. A threshold may be determined based on a system confidence metric. If the confidence metric exceeds a predetermined threshold, the system may switch to lost-in-space mode until better pose measurements are obtained from the sensors. In some embodiments, switching between lost-in-space mode and tracking mode may be performed automatically or autonomously by the system. However, in some embodiments, the switching may be performed manually by user intervention or user input. Therefore, it may be beneficial to provide a dual-mode correlation pipeline that can switch between modes depending on the available information and the associated throughput.

[0262] linear light source As previously mentioned, the active light source of a vertiport or a fiducial placed on the ground may be a point light source, such as an LED. While point light sources such as LEDs can be easy to install and modulate, in some situations it may be desirable to improve the overall signal-to-noise ratio of the optical signal generated by the ground light source. While the amount of light emitted from the light source can be increased by installing more LEDs, point light sources cannot emit light over a wide area and are therefore insufficient for high SNR applications. Therefore, systems and methods for improving the signal strength and SNR of light from a light source and received by a photodetector may be desirable.

[0263] Referring to FIG. 30 , a schematic diagram illustrates an exemplary arrangement of linear light sources in a constellation of light sources according to disclosed embodiments. The landing surface 3000 may include multiple point light sources 3030 and multiple linear light sources 3020. In some embodiments, the landing surface 3000 may include linear light sources instead of or in addition to the point light sources 3030. In some embodiments, the landing surface 3000 may be a rectangular, square, triangular, circular, or elliptical landing area, or any other suitable shape. In some embodiments, linear light sources 3020 may be disposed on all sides of the landing surface 3000 (e.g., a rectangular or square landing surface). Each side of the landing surface 3000 includes multiple collinear line segments, each including a linear light source. In some embodiments, adjacent linear light sources in the multiple line segments may be separated by point light sources. In some embodiments, no light sources may be disposed between the linear light sources in adjacent line segments, resulting in a discontinuous row of linear light sources along the side of the landing surface 3000.

[0264] The use of a linear light source can have several advantages, including, among others, a higher signal-to-noise ratio due to the greater spread of light generated from a linear source compared to a point source, more robust line detection algorithms, higher data throughput from a linear source compared to a point source, compatibility with a variety of algorithms, and simple and reliable encoding schemes.

[0265] Some advantages of using a line source in a constellation of terrestrial light sources include compatibility with detection algorithms, data association algorithms, PnP conversion, pose recovery or pose estimation algorithms, and data encoding. For example, background subtraction techniques, point detection algorithms (using point sources in conjunction with line sources), and performing line detection after background subtraction may be directly transferable, unchanged, from point source detection algorithms.

[0266] FIG. 31A is a schematic diagram of an exemplary data encoding scheme 3100 using a linear light source shown on a landing surface 3000, in accordance with disclosed embodiments. Each side can be divided into several collinear line segments without affecting the detection or association algorithms. In some embodiments, each line segment can be used to represent a single data bit. For example, a landing surface 3000 including four sides with four line segments can be configured to transmit 16 bits of data in one transmission period. Note that the number of line segments can be varied as appropriate.

[0267] An exemplary encoding scheme, such as an on / off scheme, is shown in FIG. 31A. In the on / off encoding scheme, encoding can occur by stopping the blinking of a line segment, since line detection and association work with only one segment lit on each line. Furthermore, pause recovery may require at least one line segment to blink during transmission of any binary representation. Alternatively, several fixed segments that always blink may be used for pause recovery but may be excluded from transmission. As an example, in the four-segment encoding shown in FIG. 31A, the segments closest to the corners of each line may always be toggled, allowing the inner segments to transmit data without affecting the integrity of pause recovery. The encoded binary bit shown in exemplary scheme 3100 represents the value 13. In some embodiments, the blinking of a line segment may be programmed to represent a predefined value associated with the identity of the landing surface.

[0268] 31B, an exemplary encoding scheme 3150 for data transmission using a combination of linear and point light sources is shown, according to some disclosed embodiments. An exemplary combination landing surface, e.g., landing surface 3110, can include multiple linear light sources 3120 forming the edges of a pattern, which may include, for example, a rectangle, a square, a triangle, etc., and multiple point light sources 3125 distributed within the area bounded by the linear light sources 3120. In some embodiments, the point light sources 3125 may be distributed in a predetermined pattern with known positions, or in a randomly generated pattern where the positions of the point light sources are unknown.

[0269] In some embodiments, data transmission between a landing surface and an airborne vehicle configured to land or take off from the landing surface may include providing an encoding scheme. The encoding scheme 3150 may include labeling intersections of linear light sources 3120. The labeling scheme may include labeling points in a predetermined, known order using heading information from the INS or one or more previous iterations. For example, in the rectangular pattern shown in FIG. 31B, northeast may be labeled "0," southeast may be labeled "1," southwest may be labeled "2," and northwest may be labeled "3." As another example, intersections may be labeled based on the number of nearby point light sources. The corner with the fewest nearby point light sources may be labeled "0," and the corner with the most nearby point light sources may be labeled "3." While only two labeling schemes are discussed, it should be understood that other suitable labeling schemes may also be applied.

[0270] The data transmission may further include projecting the light source onto a normalized grid 3130. One method of projecting the light source includes calculating and applying a homography matrix to remove distortion associated with the projection. The normalized grid 3130 may be divided into a predetermined number of subspaces 3140. While the normalized grid 3130 is shown divided into nine subspaces 3140, the normalized grid 3130 may be divided into any number of subspaces based on the number of point sources, the density of the point sources, the area of ​​the point sources, or as needed. For each subspace, if a point source is detected (i.e., active or on), it may be labeled "1," and if no point source is detected, it may be labeled "0." The labels of each subspace may be combined in a predetermined order to generate a binary value configured to represent or identify a landing space. By way of example, as shown in the normalized grid 3130, the labels of the nine subspaces 3140 may be combined to form the binary value 100010011, which represents the number 275. In some embodiments, the number 275 represents, among other things, identification information associated with the landing surface, or the spatial orientation of the landing surface, or an authentication code that can be used to verify the identity of a user or the identity of an aircraft.

[0271] 31C shows a flowchart of an example method 3160 for pose estimation using a line light source, consistent with certain disclosed embodiments. Method 3160 may be performed in combination with or instead of methods for pose estimation using a point light source (as described above).

[0272] Method 3160 includes, but is not limited to, receiving at least two images from a camera mounted on an aerial vehicle, performing background subtraction from the received images, performing line detection using a line detection algorithm such as a Hough transform or a line detection filter, identifying where the lines intersect, labeling each intersection using a labeling scheme, running a PnP transformation algorithm using the labeled points, and selecting a valid pose from the results. Steps represented by the shadow blocks, such as background subtraction, connected components, centroiding, association, and PnP, are processes present in the point light source pose estimation method. One or more steps for point light source pose estimation may be used in addition to or instead of pose estimation using a line light source.

[0273] In some embodiments, the labeling scheme may include labeling points in a predetermined, known order using heading information from the INS or one or more previous iterations. For example, in the rectangular pattern shown in FIG. 31B, northeast may be labeled "0," southeast may be labeled "1," southwest may be labeled "2," and northwest may be labeled "3." As another example, intersections may be labeled based on the number of nearby point sources. The corner with the fewest nearby point sources may be labeled "0," and the corner with the most nearby point sources may be labeled "3."

[0274] Data Augmentation In urban environments, GPS signals can be delayed, obscured, distorted, or completely undetectable due to reflections and obstructions from densely packed structures such as tall buildings and towers. GPS signals may be reflected by buildings, walls, vehicles, and even the ground. Glass, metal, and wet surfaces are known to be strong reflectors of light. These reflected signals can interfere with reception of signals received directly from the satellites. For example, signals may be received via multiple paths due to reflections from other surfaces and structures near the aircraft, a phenomenon known as multipath interference or multipath effect, as shown in FIG. 32B. However, in some cases, direct signals from GPS satellites (e.g., GPS satellite 3220) may be blocked or obstructed by tall buildings 3210, and only reflected signals may be received, a phenomenon known as non-line-of-sight (NLOS) reception, as shown in FIG. 32A. In the context of this disclosure, multipath effects pose several challenges and can be more troublesome than NLOS reception because measurements from GPS signals are not simply distorted but can also become undetectable. In some cases, a single signal may be received twice or the signal may be significantly delayed, which can directly impact the time-of-arrival (TOA) calculations required to generate a position. These effects may be accentuated at landing and takeoff locations where the airborne vehicle approaches a landing surface and is at an elevation where signals may be obstructed, reflected, or distorted by surrounding structures. Proposed precision landing and takeoff (PLaTO) systems and methods, as discussed in several embodiments herein, may be used to eliminate one or more of these effects to improve the overall accuracy of position measurements.

[0275] a. Use of GPS to augment PLato As previously described, the algorithms used in PLaTO may be iterative in nature, and as a result, providing an initial estimate may reduce convergence time and signal latency. In some existing systems and methods, if an initial estimate is not available, it may take milliseconds to seconds to execute one or more steps of the algorithm, making the algorithm inefficient and negatively impacting data throughput. However, in some embodiments, GPS signals may be used to seed the algorithm with an initial guess for the aircraft's current location, thereby reducing commute times during "lost in space" mode. Even an inaccurate estimate may contribute to a faster algorithm by reducing the number of iterations required to achieve convergence.

[0276] 33, there is shown a data augmentation pipeline 3300 configured to combine GPS and Inertial Navigation System (INS) measurements to improve the accuracy or speed of aircraft-level positioning, in accordance with disclosed embodiments. In practice, the aircraft may be configured to use some form of GPS positioning, and information from the INS may also be available.

[0277] The data augmentation pipeline 3300 may include receiving information related to the position of the aircraft based on GPS signals from one or more GPS satellites. In some embodiments, the position information may include the position coordinates of the aircraft. The pipeline 3300 may further include receiving information related to the position of a landing surface or vertiport, which may already exist in a database. In some embodiments, the position of the landing surface may be originally determined based on GPS signals and stored in an accessible database for later use. The pipeline 3300 may further include receiving information related to the attitude (pose, position, orientation) of the aircraft based on INS measurements. Using information related to the position of the aircraft in 3D space, the position of the landing surface in 3D space, and the orientation of a camera configured to determine the attitude of the aircraft, it may be possible to determine a region of interest 3350. In this context, "region of interest" refers to an area allocated as a landing space or vertiport for an eVTOL aircraft. In some embodiments, to reduce processing time, region 3355 (shown by the pixelated area around region of interest 3350) may be excluded from consideration. In some embodiments, as previously mentioned, the GPS information relating to the aircraft's location may not need to be precise, and even an initial guess may be useful to reduce processing time.

[0278] b. Using PLaTO to augment GPS measurements As shown in FIG. 34, a data augmentation pipeline 3400 configured to augment GPS measurements with information from a PLaTO system is shown in accordance with a disclosed embodiment.

[0279] As previously mentioned, multipath effects in dense urban environments pose several challenges associated with receiving GPS signals and adversely affect the accuracy of position determinations. In some embodiments, the PLaTO system may be used to help mitigate some of these issues in order to improve the overall accuracy of GPS position determinations.

[0280] In some embodiments, the pipeline 3400 may include using position measurements from a PLaTO system to constrain the position estimate provided by the GPS signal. In some cases, the GPS signal may be affected by multipath effects, NLOS reception, or both, such that the GPS signal may only provide a position estimate but not an accurate measurement. At least five satellites may be desirable for a reliable GPS signal. In scenarios with more than five satellites, one or more satellites may be rejected if they result in a position reference outside the range of the PLaTO estimate. As shown in FIG. 34, if the number of detected satellites (N) is less than the minimum number of satellites required to generate a reliable GPS signal, one or more satellites may be verified. If the number of satellites exceeds the minimum number of satellites, one or more satellites may be rejected and the GPS signal may be calculated.

[0281] In some embodiments, if GPS satellites cannot be verified, they may be deemed unable to provide a position estimate or may be deemed poorly performing satellites and removed from consideration. A position may then be calculated using the remaining satellites. In some embodiments, verifying existing satellites may include estimating GPS signals from available GPS satellites, determining for each satellite whether the error between the PLaTO signal and the GPS signal is below a predefined threshold error limit, and then calculating a position based on the signal received from the satellite. If the error exceeds the predefined threshold limit, the signal from the satellite and the satellite may be removed from consideration.

[0282] In some embodiments, specific position measurements from the PLaTO can identify the positions of individual satellites. Information related to the positions of individual satellites may be used to remove multipath effects or combine satellite signals. To remove multipath effects, one or more of the following steps may be performed: Providing satellite measurements such as Time Difference of Arrival (TDOA) and position measurements from the i.PLaTO system. ii. Calculating the expected TDOA signal for each satellite based on the position reported by the PLaTO system. iii. Comparing the measured TDOA signal to the expected TDOA signal and filtering out outliers that fall outside the specified range. iv. Calculating position using one or more expected TDOAs, allowing GPS to calculate position with fewer satellites than would normally be required.

[0283] In some embodiments, based on position measurements from PLaTO and less than a minimum number of GPS satellites (e.g., two satellites), the position probability distribution of the two measurements can be combined with a Kalman filter to reduce the measurement error beyond what either system could produce individually.

[0284] Use of PLaTO to augment c.INS measurements In the context of this disclosure, using the last known position and extrapolating based on INS measurements is known as dead reckoning navigation. While this method can be useful for short-distance travel, for longer distances, INS sensor drift can result in the accumulation of large errors over time. This can be because the INS measures acceleration directly and generates velocity and position estimates by integrating acceleration over time. Therefore, small offsets in acceleration can accumulate into large position errors over time. In some embodiments, measurements from PLaTO can be used to augment the INS measurements, as shown in FIG. 35.

[0285] In some embodiments, absolute corrections of the aircraft's position over time may be obtained from the PLaTO system. This may be achieved by fusing position and velocity estimates from the PLaTO system with INS measurements to produce highly accurate measurements. One or more camera images from the PLaTO system can be used to perform visual odometry to augment the INS measurements. In the context of this disclosure, "visual odometry" is similar to the point tracking described in the data association algorithm above, but the end result is a change in position between frames rather than a position. One of the advantages of visual odometry is that it does not require a fiducial to be present in the frame to function, so it can be used even when no landing surface or vertiport is nearby during normal flight. Visual odometry algorithms: 1. Time t a and t b providing the image (a) and the image (b) captured in the 2. Identifying and distinguishing features of image (a) and image (b); 3. Associating features of image (a) with features of image (b) using a nearest neighbor method or suitable algorithm; 4. Estimating the change in camera position between image (a) and image (b).

[0286] In some embodiments, the results of the visual odometry algorithm are calculated based on the time at which the image was captured. a and t b This may be used to correct for drift that occurs in the time range between t a and t b The difference between them is 1, so t a and t b The images captured by the camera are sequential. Changes in position may be similarly integrated over time to produce a relative position from a starting point. Another advantage of visual odometry is that it is less susceptible to drift effects, since it is tied to the visual range of the camera and the environment rather than to acceleration measurements.

[0287] d. INS assisted by optical localization system Referring to FIG. 36 , an exemplary pipeline 3600 of a data augmentation method according to some disclosed embodiments is shown. Pipeline 3600 illustrates a method for using a fixed-delay smoother algorithm to augment INS measurements with an optical localization system. In some embodiments, an optical localization system using a fixed-delay smoother algorithm may be used on an aircraft in combination with an INS to provide a more accurate position estimate than either used individually. A standard filtering implementation utilizes an extended Kalman filter (EKF) to fuse the optical localization pose solution with inertial navigation system data. However, as an alternative to an EKF, a pose graph optimization approach, such as a fixed-delay smoother, may be used to calculate an optimal position estimate using sensor data over an entire window of time, rather than just instantaneous time. In some embodiments, using a fixed-delay smoother algorithm may improve position estimation accuracy at high frequencies, such as 500 Hz.

[0288] e. PLaTO integration with aircraft Referring to FIG. 37 , an exemplary system 3700 illustrating the integration of a PLaTO system with an aircraft to support piloted or unmanned flight is shown, according to some disclosed embodiments. The electric propulsion system of an eVTOL may include an electric engine that provides mechanical shaft power to a propeller assembly to generate thrust. In some embodiments, the electric engine of the electric propulsion system may include a high-voltage power system that supplies high-voltage electrical power to the electric engine and / or a low-voltage system that supplies low-voltage DC electrical power to the electric engine. Some embodiments may include the electric engine(s) in digital communication with a flight control system (“FCS”) that includes a flight control computer (“FCC”) 3750 that can send and receive signals to the electric engine, including commands and response data or status. Some embodiments may include the electric engine that can receive operating parameters from the FCC, including values ​​of speed, voltage, current, torque, temperature, vibration, propeller position, and other operating parameters, and communicate the operating parameters to the FCC.

[0289] In some embodiments, the flight control system may include a system that communicates with the electric engines to send and receive analog / discrete signals to the electric engines and control devices that can redirect the thrust of the tilt propellers from a primarily vertical orientation during vertical flight mode to a primarily horizontal orientation during forward flight mode. In some embodiments, this system is referred to as a tilt propeller system ("TPS"), and can communicate with and orient additional functions of the electric propulsion system.

[0290] In some embodiments, the system 3700 may communicate the measured pose (position and orientation) to the FCC 3750. In some embodiments, the FCC 3750 may fuse estimated pose from other sources, such as the GPS 3710, INS 3720, altimeter 3710, and PLaTO 3740, to generate an optimal estimate of the aircraft pose. This may be performed using several sensor fusion techniques, such as a Kalman filter, an extended Kalman filter, a fixed-lag smoother, or other methods for performing sensor fusion.

[0291] In piloted aircraft, the final position estimate of the aircraft may be used to provide visual feedback to the pilot. In unmanned aerial vehicles, the final position estimate may be used to calculate flight control commands such as motor commands, flight surface controls 3770, or other control signals used to pilot the aircraft during flight.

[0292] Example - EKF using position data from PLaTO Figures 38-41 show real-time flight test results for an extended Kalman filter test using position data from the PLaTO system. Figure 38 shows a comparison of altitude estimates plotted as a function of horizontal distance measured by RTK-GPS ground truth data and EKF. As shown in Figure 38, the EKF line (dotted) primarily follows the RTK line (solid). Figure 39 shows a comparison of the aircraft's ground altitude measured by EKF and RTK-GPS ground truth data and the corresponding error. As shown in Figure 39, the EKF line (dotted) primarily follows the RTK line (solid). Figure 40 shows the east estimate and its corresponding error plot as a function of ground truth. As shown in Figure 40, the EKF line (dotted) primarily follows the RTK line (solid). Figure 41 shows the north estimate and corresponding error plot as a function of ground truth. As shown in Figure 41, the EKF line (dotted) primarily follows the RTK line (solid).

[0293] Referring to FIG. 42, a flowchart illustrates an exemplary method 4200 for estimating the pose of an airborne vehicle in accordance with an embodiment of the present disclosure. The method 4200 may be implemented in computing devices and systems such as those disclosed herein. In some embodiments, the method 4200 may be performed by at least one processor of a computer-implemented system. The steps and methods of each of these components of the method 4200 are described below. It will be understood that the components and methods may be combined, modified, or rearranged depending on the application and system embodiment.

[0294] As shown in FIG. 42, in step 4210, a landing surface may be provided that includes light sources arranged in a predetermined pattern. The landing surface may be a vertiport for an eVTOL airborne vehicle. Each light source may be an active light source configured to emit light, and one or more characteristics of the emitted light may be modulated over time. The light sources may be arranged in a predetermined pattern, with each light source having a known location. A constellation design of the arranged light sources may include an arrangement of light sources that defines a set of intersecting imaginary lines, with a light source arranged on each imaginary line and with unequal distances between adjacent light sources on each imaginary line.

[0295] In step 4220, one or more characteristics of a light source on the landing surface may be modulated over time. The light source characteristics may include the intensity, frequency, amplitude, wavelength, phase, bandwidth, or duty cycle of the emitted light. Modulation of one or more characteristics of the emitted light may identify the landing surface, identify the light source, identify the location of the light source, identify the operational state of the landing surface, or encode a signal that authenticates the landing surface. The light source characteristics may be modulated by a controller on the landing surface.

[0296] In step 4230, a camera mounted on the airborne vehicle may receive an input signal related to light emitted from the light source. The camera may be mounted on the airborne vehicle at a known position and a known orientation. The camera may use an optical filter or lens to allow a range of wavelengths.

[0297] In step 4240, based on the received input signals, the camera may generate output in the form of a still image, a series of still images, or streaming video of information captured from the landing surface and its surroundings.

[0298] In step 4250, the processor may determine a position and orientation of the airborne vehicle based on information in the image captured by the camera. Determining the position and orientation of the airborne vehicle may include detecting at least one light source in the image, determining which of the light sources arranged in a predetermined pattern the detected light source is, and determining the position and orientation of the airborne vehicle based on the determination of which of the light sources arranged in the predetermined pattern the detected light source is. The processor may be configured to execute one or more algorithms to estimate a pose of the airborne vehicle based on information received from the camera.

[0299] FIG. 44 illustrates a perspective view of an exemplary VTOL aircraft according to disclosed embodiments. FIG. 45 illustrates another example of a perspective view of an exemplary VTOL aircraft in an alternative configuration according to embodiments of the present disclosure. FIGS. 44 and 45 illustrate VTOL aircraft 4400, 4500, respectively, in a cruise configuration and a vertical takeoff and landing and hover configuration (also referred to herein as a "lift" configuration) according to embodiments of the present disclosure. Corresponding elements in FIGS. 44 and 45 may have similar numerals and refer to similar elements of the aircraft 4400, 4500. The aircraft 4400, 4500 may include a fuselage 4402, 4502, wings 4404, 4504 attached to the fuselage 4402, 4502, and one or more aft stabilizers 4406, 4506 attached to the aft of the fuselage 4402, 4502. Multiple lift propellers 4412, 4512 may be attached to the wings 4404, 4504 and configured to provide lift for vertical takeoff and landing and hovering. Multiple tilt propellers 4414, 4514 may be attached to the wings 4404, 4504 and can be tilted between a lift configuration, which provides a portion of the lift required for vertical takeoff, landing, and hovering, as shown in FIGURE 45, and a cruise configuration, which provides forward thrust to the aircraft 4400 for horizontal flight, as shown in FIGURE 44. As used herein, tilt propeller lift configuration refers to any tilt propeller orientation in which the tilt propeller thrust is primarily providing lift to the aircraft, and tilt propeller cruise configuration refers to any tilt propeller orientation in which the tilt propeller thrust is primarily providing forward thrust to the aircraft.

[0300] In some embodiments, the lift propellers 4412, 4512 may be configured to provide lift only, with all horizontal propulsion provided by the tilt propellers. Thus, the lift propellers 4412, 4512 may be configured in a fixed position and generate thrust only during takeoff, landing, and hovering flight phases. Meanwhile, the tilt propellers 4414, 4514 may be tilted upward into a lift configuration in which thrust is directed downward, providing additional lift.

[0301] For forward flight, the tilt propellers 4414, 4514 may tilt from a lift configuration to a cruise configuration. In other words, the orientation of the tilt propellers 4414, 4514 may change from an orientation in which the tilt propeller thrust is directed downward (to provide lift during vertical takeoff and landing and hovering) to an orientation in which the tilt propeller thrust is directed rearward (to provide forward thrust for the aircraft 4400, 4500). The tilt propeller assembly for a particular electric engine may tilt about an axis of rotation defined by the attachment point connecting the boom and the electric engine. When the aircraft 4400, 4500 is in full forward flight, lift may be provided entirely by the wings 4404, 4504. Meanwhile, in the cruise configuration, the lift propellers 4412, 4512 may be turned off. The blades 4420, 4520 of the lift propellers 4412, 4512 may be held in a low-drag position when the aircraft is cruising. In some embodiments, the lift propellers 4412, 4512 each have two blades 4420, 4520 that can be locked for cruising in a minimum-drag position with one blade positioned directly above the other, as shown in FIG. 474. In some embodiments, the lift propellers 4412, 4512 can have more than two blades. In some embodiments, the tilt propellers 4414, 4514 may include more blades 4416, 4516 than the lift propellers 4412, 4512. For example, as shown in FIGS. 44 and 45, the lift propellers 4412, 4512 can each include, for example, two blades, while the tilt propellers 4414, 4514 may each include more blades, such as five blades as shown. In some embodiments, each tilt propeller 4414, 4514 may have between 2 and 5 blades, and possibly more depending on the design considerations and requirements of the aircraft.

[0302] In some embodiments, the aircraft may include a single wing 4404, 4504 on each side of the fuselage 4402, 4502 (or a single wing across the entire aircraft). At least a portion of the lift propellers 4412, 4512 may be located aft of the wings 4404, 4504, and at least a portion of the tilt propellers 4414, 4514 may be located forward of the wings 4404, 4504. In some embodiments, all of the lift propellers 4412, 4512 may be located aft of the wings 4404, 4504, and all of the tilt propellers 4414, 4514 may be located forward of the wings 4404, 4504. According to some embodiments, all of the lift propellers 4412, 4512 and tilt propellers 4414, 4514 may be mounted on the wings, i.e., no lift or tilt propellers may be mounted on the fuselage. In some embodiments, the lift propellers 4412, 4512 may all be located aft of the wings 4404, 4504, and the tilt propellers 4414, 4514 may all be located forward of the wings 4404, 4504. According to some embodiments, all of the lift propellers 4412, 4512 and tilt propellers 4414, 4514 may be located inboard of the ends of the wings 4404, 4504.

[0303] In some embodiments, the lift propellers 4412, 4512 and tilt propellers 4414, 4514 may be attached to the wings 4404, 4504 by booms 4422, 4522. The booms 4422, 4522 may be below the wings 4404, 4504, above the wings, and / or integrated into the wing profile. In some embodiments, the lift propellers 4412, 4512 and tilt propellers 4414, 4514 may be attached directly to the wings 4404, 4504. In some embodiments, the lift propellers 4412, 4512 and tilt propellers 4414, 4514 may be attached to the booms 4422, 4522, respectively. The lift propellers 4412, 4512 may be attached to the aft ends of the booms 4422, 4522, and the tilt propellers 4414, 4514 may be attached to the forward ends of the booms 4422, 4522. In some embodiments, the lift propellers 4412, 4512 may be attached to fixed locations on the booms 4422, 4522. In some embodiments, the tilt propellers 4414, 4514 may be attached to the forward ends of the booms 4422, 4522 via hinges. The tilt propellers 4414, 4514 may be attached to the booms 4422, 4522 so as to be aligned with the main body of the booms 4422, 4522 in the cruise configuration, forming a continuous extension of the forward ends of the booms 4422, 4522 that minimizes drag in forward flight.

[0304] In some embodiments, the aircraft 4400, 4500 may include, for example, one wing on each side of the fuselage 4402, 4502, or a single wing spanning the entire aircraft. According to some embodiments, at least one wing 4404, 4504 is a high wing attached to the upper side of the fuselage 4402, 4502. According to some embodiments, the wing includes control surfaces such as flaps and ailerons. In some embodiments, the wing 4404, 4504 may be designed to reduce drag during forward flight. In some embodiments, the wing tip profile may be curved or tapered to minimize drag.

[0305] In some embodiments, the aft stabilizer 4406, 4506 includes control surfaces such as one or more rudders, one or more elevators, and / or one or more rudder-elevator combinations. The wings can have any suitable design. In some embodiments, the wings have tapered tips.

[0306] In some embodiments, a lift propeller 4412, 4512 or tilt propeller 4414, 4514 may be tilted relative to at least one other lift propeller 4412, 4512 or tilt propeller 4414, 4514. As used herein, "tilt" refers to the relative orientation of the lift / tilt propeller's rotational axis about a line parallel to the longitudinal direction, analogous to an aircraft's roll degree of freedom. Lift and / or tilt propeller tilt can help minimize damage from propeller rupture and may improve yaw control during flight by adjusting the plane of rotation of the lift / tilt propeller disk (the blades and hub to which the blades are attached) so that it does not intersect with critical parts of the aircraft (such as parts of the fuselage where personnel may be located, critical flight control systems, batteries, adjacent propellers, etc.) or other propeller disks.

[0307] FIG. 46 illustrates a top view of an exemplary VTOL aircraft according to an embodiment of the present disclosure. The aircraft 4600 illustrated in the figure may be a top view of the aircraft 4400, 4500 illustrated in FIGS. 44 and 45. As described herein, the aircraft 4600 may include twelve electric propulsion systems distributed throughout the aircraft 4600. In some embodiments, the distribution of the electric propulsion systems may include six forward electric propulsion systems 4614 and six aft electric propulsion systems 4612 mounted on forward and aft booms of the main wing 4604 of the aircraft 4600. In some embodiments, the length of the aft end of the boom 4624 from the wing 4604 to the lift propeller 4612 may be similar to the length of the aft end of the boom 4624 across the multiple aft ends of the boom. In some embodiments, the length of the aft end of the boom may vary across the six aft ends of the exemplary boom. For example, each aft section of the boom 4624 may have a different length from the wing 4604 to the lift propeller 4612, or a subset of the aft sections of the boom may be similar in length. In some embodiments, the front ends of the booms 4622 may include a variety of lengths from the wing 4604 to the tilt propeller 4614 across the front ends of the booms. For example, as shown in FIG. 46 , the length of the front end of the boom 4622 from the tilt propeller 4614 closest to the fuselage to the wing 4604 may be greater than the length of the front end of the boom 4622 to the tilt propeller 4614 farthest from the fuselage. Some embodiments may have similar lengths across the six front ends of the exemplary booms from the wing 4604 to the tilt propeller 4614, or any other distribution of lengths for the front ends of the booms. Some embodiments may include an aircraft 4600 with four forward electric propulsion systems 4614 and four aft electric propulsion systems 4612, for a total of eight electric propulsion systems, or any other distribution of forward and aft electric propulsion systems, including embodiments in which the number of forward electric propulsion systems 4614 is less than or greater than the number of aft electric propulsion systems 4612. Additionally, FIG. 46 illustrates an exemplary embodiment of a VTOL aircraft 4600 with forward propellers 4614 positioned horizontally for level flight and aft propeller blades 4620 in a stowed position for the forward phase of flight.

[0308] As disclosed herein, the forward and aft electric propulsion systems may be either clockwise (CW) or counterclockwise (CCW) types. Some embodiments may include various forward electric propulsion systems that are a mix of both CW and CCW types. In some embodiments, the aft electric propulsion system may comprise a mix of CW and CCW type systems among the aft electric propulsion systems.

[0309] FIG. 47 is a schematic diagram illustrating exemplary propeller rotation for a VTOL aircraft according to disclosed embodiments. The aircraft 4700 shown may be a top view of the aircraft 4400, 4500, and 4600 shown in FIGS. 1, 2, and 3. The aircraft 4700 may include six forward electric propulsion systems, three of which are CW type 4724 and the remaining three are CCW type. In some embodiments, the three aft electric propulsion systems may be CCW type 4728 and the remaining three aft electric propulsion systems may be CW type 4730. Some embodiments may include the aircraft 4700 with four forward electric propulsion systems and four aft electric propulsion systems, two of which are CW type and two of which are CCW type. In some embodiments, the propellers may counter-rotate relative to adjacent propellers to cancel torque steer induced on the fuselage or wings by the propeller rotation. In some embodiments, the difference in rotation direction may be achieved using the rotation direction of the engines. In other embodiments, the engines may all rotate in the same direction and gearing may be used to achieve different propeller rotation directions.

[0310] Some embodiments may include an aircraft 4700 with forward and aft electric propulsion systems, where the amount of CW type 4724 and CCW type 4726 is unequal between the forward electric propulsion system, the aft electric propulsion system, or the forward and aft electric propulsion systems.

[0311] FIG. 48 is a schematic diagram illustrating exemplary power connections in a VTOL aircraft according to disclosed embodiments. The VTOL aircraft may have various power systems connected to diagonally opposed electric propulsion systems. In some embodiments, the power systems may include high-voltage power systems. Some embodiments may include high-voltage power systems connected to electric engines via high-voltage channels. In some embodiments, the aircraft 4800 may include six power systems including batteries 4826, 4828, 4830, 4832, 4834, and 4836 housed within wings 4870 of the aircraft 4800. In some embodiments, the aircraft 4800 may include six forward electric propulsion systems having six electric engines 4802, 4804, 4806, 4808, 4810, and 4812 and six aft electric propulsion systems having six electric engines 4814, 4816, 4818, 4820, 4822, and 4824. In some embodiments, the batteries may be connected to the diagonally opposed electric engines. In this configuration, first power supply system 4826 can provide power to electric engine 4802 via power connection channel 4838 and to electric engine 4824 via power connection channel 4840. In some embodiments, first power supply system 4826 can also be paired with fourth power supply system 4832 via power connection channel 4842 that includes a fuse to prevent excessive current from flowing through power supply systems 4826 and 4832. Further to this embodiment, VTOL air vehicle 4800 can include second power supply system 4828 paired with fifth power supply system 4834 via power connection channel 4848 that includes a fuse and can provide power to electric engines 4810 and 4816 via power connection channels 4844 and 4846, respectively. In some embodiments, third power supply system 4830 can be paired with sixth power supply system 4836 via power connection channel 4854 that includes a fuse and can provide power to electric engines 4806 and 4820 via power connection channels 4850 and 4852, respectively.Fourth power system 4832 may also supply power to electric engines 4808 and 4818 via power connection channels 4856 and 4858, respectively. Fifth power system 4834 may also supply power to electric engines 4804 and 4822 via power connection channels 4860 and 4862, respectively. Sixth power system 4836 may also supply power to electric engines 4812 and 4814 via power connection channels 4864 and 4866, respectively.

[0312] As disclosed herein, an electric propulsion system may include an electric engine connected to a high-voltage power supply system, such as batteries, located within the aircraft via a high-voltage channel or power connection channel. In some embodiments, various batteries are housed within the aircraft's wings, and high-voltage channels are connected to the electric propulsion system throughout the aircraft, including the wings and boom. In some embodiments, multiple high-voltage power supply systems can be used to create an electric propulsion system with multiple high-voltage power sources to avoid the risk of a single point of failure. In some embodiments, an aircraft may include multiple electric propulsion systems that may be wired to various batteries or power sources stored throughout the aircraft. It is recognized that such a configuration can be beneficial to avoid the risk of a single point of failure, where the failure of one battery or power supply could prevent a portion of the aircraft from maintaining the required thrust and thus preventing it from continuing flight or performing a controlled landing. For example, if a VTOL aircraft includes two forward electric propulsion systems and two aft electric propulsion systems, the forward and aft electric propulsion systems on opposite sides of the VTOL aircraft may be connected to the same high-voltage power supply system. In such a configuration, if one high-voltage power supply system fails, the diagonally opposite forward and aft electric propulsion systems of the VTOL aircraft continue to operate, potentially providing a more balanced flight or landing compared to a case where the forward and aft electric propulsion systems fail on the same side of the VTOL aircraft. Some embodiments may include four forward electric propulsion systems and four aft electric propulsion systems, with the diagonally opposite electric engines connected to a common battery or power source. Some embodiments may include various configurations of electric engines electrically connected to the high-voltage power supply systems, thereby avoiding the risk of a single point of failure in the event of a power supply failure, allowing the phase of flight in which the failure occurs to continue or the aircraft to perform an alternate phase of flight in response to the failure.

[0313] As described above, an electric propulsion system may include an electric engine that provides mechanical shaft power to a propeller assembly to generate thrust. In some embodiments, the electric engine of an electric propulsion system may include a high-voltage power system that supplies high-voltage electrical power to the electric engine and / or a low-voltage system that supplies low-voltage DC electrical power to the electric engine. Some embodiments may include the electric engine(s) in digital communication with a flight control system (“FCS”) that includes a flight control computer (“FCC”) that can send and receive signals to the electric engine, including commands and response data or status. Some embodiments may include the electric engine that can receive operating parameters from the FCC, including values ​​of speed, voltage, current, torque, temperature, vibration, propeller position, and other operating parameters, and communicate the operating parameters to the FCC.

[0314] In some embodiments, the flight control system may include a system that communicates with the electric engines to send and receive analog / discrete signals to the electric engines and control devices that can redirect the thrust of the tilt propellers from a primarily vertical orientation during vertical flight mode to a primarily horizontal orientation during forward flight mode. In some embodiments, this system may be referred to as a tilt propeller system ("TPS"), and can communicate with and orient additional functions of the electric propulsion system.

[0315] FIG. 49 illustrates a block diagram of an exemplary architecture and design of an electric propulsion unit 4900 according to disclosed embodiments. In some embodiments, the electric propulsion system 4902 may include an electric engine subsystem 4904 that may provide torque via a shaft to a propeller subsystem 4906, generating thrust for the electric propulsion system 4902. Some embodiments may include the electric engine subsystem 4904 receiving low-voltage DC (LVDC) power from a low-voltage system (LVS) 4908. Some embodiments may include the electric engine subsystem 4904 receiving high-voltage (HV) power from a high-voltage power system (HVPS) 4910 that includes at least one battery or other device capable of storing energy. In some embodiments, the high-voltage power system may include multiple batteries or other devices capable of storing energy that provide high-voltage power to the electric engine subsystem 4904. It is recognized that such a configuration may be advantageous because it does not risk a single point of failure, where failure of a single battery could lead to failure of the electric propulsion system 4902.

[0316] Some embodiments may include an electric propulsion system 4902 including an electric engine subsystem 4904 that receives signals from and transmits signals to a flight control system 4912. In some embodiments, the flight control system 4912 may include a flight control computer that can send commands to and receive status and data from the electric engine subsystem 4904 using controller area network (“CAN”) data bus signals. While CAN data bus signals are used between the flight control computer and the electric engine(s), it should be understood that some embodiments may include any form of communication capable of sending and receiving data from the flight control computer to the electric engine(s). In some embodiments, the flight control system 4912 may also include a tilt propeller system (“TPS”) 4914 that can send and receive analog and discrete data to and from the tilt propeller electric engine subsystem 4904. The tilt propeller system 4914 may include a device that communicates operating parameters to the electric engine subsystem 4904 and that can adjust the orientation of the propeller subsystem 4906 to redirect the thrust of the tilt propeller during various phases of flight using mechanical means (e.g., a gearbox assembly, linear actuators, or other components) to change the orientation of the propeller subsystem 4906.

[0317] As discussed, exemplary VTOL aircraft may be equipped with various types of electric propulsion systems with tilt and lift propellers, including forward electric engines with the ability to tilt during various flight phases, and aft electric engines that remain in one direction and may only be active during certain flight phases (i.e., takeoff, landing, hovering).

[0318] In some embodiments, the flight control system may include a system capable of controlling control surfaces and associated actuators in an exemplary VTOL aircraft. FIG. 50 illustrates a top view of an exemplary VTOL aircraft 5000 according to an embodiment of the present disclosure. The aircraft 5000 illustrated in the figure may be a top view of the aircraft 4400, 4500 illustrated in FIGS. 44 and 45. In some embodiments, the aircraft 5000 may resemble the aircraft 4600 illustrated in FIG. 46. In the aircraft 5000, the control surfaces may include, in addition to the propeller blades previously described, a flaperon 5072 and a ruddervator 5074. The flaperon 5072 may combine the functionality of one or more flaps, one or more ailerons, and / or one or more spoilers. The ruddervator 5074 may combine the functionality of one or more rudders and / or one or more elevators. In some embodiments, the control surfaces may include, for example, a flap, an aileron, a spoiler, a rudder, or an elevator. In the aircraft 5000, the actuators may include control surface actuators (CSAs) associated with the flaperons 5072 and ruddervators 5074 in addition to the electric propulsion system described above.

[0319] Example - Using IR Random Dot Markers for Landing Random-dot markers are proven to be robust to occlusions and reliable. Rather than relying solely on frame-by-frame point tracking to identify points at shallow viewing angles, the algorithm used in this example employs improved nearest neighbor and descriptor calculations to allow markers to be redetected even if tracking fails.

[0320] Light detection: The light source flashes at a frequency 1 / 3 the camera's frame rate, and the frames are processed in batches of three. The maximum and minimum grayscale intensities of each pixel across the three images are determined, and a maximum and minimum image are constructed. The minimum image is then subtracted from the maximum image to remove ambient background infrared light. In the Min-Max image, the constellation appears in high contrast against the background, and the pixel locations of the light can be used for both keypoint registration and acquisition.

[0321] Keypoint Registration: Before live pose estimation can occur, the constellation must be registered using known light positions and IDs. To identify points using LLAH, multiple "descriptors" are calculated for each keypoint. To calculate the descriptors, LLAH finds the n-neighbors of each keypoint. In this step, the constellation is rescaled to have a 1:1 length-to-width ratio; otherwise, the nearest neighbors may change under drastic viewpoint transformations. Because the crossover ratio is order-dependent, the nearest neighbors are sorted according to their clockwise position relative to the keypoint. Next, m-point combinations are selected from the sorted neighborhood. From those m points, a four-point combination and the keypoint are used to calculate the crossover ratio. The crossover ratio for each four-point combination is calculated, discretized, and saved in the calculated order. Finally, the descriptors are saved, preserving the original order, starting with the lowest discretized crossover ratio value in the sequence. Overall, each descriptor has a dimension of mC4, and each point has nCm descriptors. A hash index is calculated from each descriptor, and the keypoint ID is saved along with the descriptor associated with this index.

[0322] Keypoint Acquisition: During live processing, markers can be acquired using either matching or tracking. If all keypoints are identified in the previous frame, the algorithm defaults to basic point tracking between the previous and current frames. Otherwise, the algorithm attempts both matching and tracking and uses the result of the method that identifies the most points. To acquire markers using matching, a descriptor is calculated for each detected point in the image using the same method as for registration. The descriptors can then be used to calculate a hash index. In this index, a vote is taken for each keypoint ID candidate with a matching descriptor. A keypoint is identified by the ID with the most votes above a certain threshold, unless the ID has already been used to identify another point. After all keypoints have been processed during the matching process, a homography is calculated between the live frame and points from the known constellation to confirm matches and identify points that were not assigned an ID during matching.

[0323] Pose Estimation: Once at least 20 of the 25 points are correctly identified in a frame, the aircraft's pose relative to the constellation is determined using OpenCV's iterative viewpoint-point (PNP) pose estimation. A pose estimate is accepted if the change in roll between the previous and current frame does not exceed 2 degrees. Adding this constraint ensures that impossible PNP poses are rejected because the aircraft cannot rotate twice in a single frame's time.

[0324] Results: Live data was acquired using a helicopter approaching a randomly placed constellation during bright daylight hours. Detecting infrared LEDs during the day can be more difficult than at night due to clutter from external reflections. The algorithm was able to identify the light at a distance of approximately 200 meters from the constellation. The current Python implementation of this data association algorithm takes 923 milliseconds to acquire a frame using matching and 0.371 milliseconds to acquire a frame using tracking. In this dataset, 21% of the frames were acquired purely through tracking, while the remainder required the matching process. As shown in Figure 51, the average position error in each direction was -1.46 m, -0.41 m, and -0.39 m, with standard deviations of 0.35 m, 0.34 m, and 0.22 m, respectively.

[0325] Embodiments of the present disclosure may be further described with respect to the following clauses. Clause 1. A system comprising: a landing surface for an airborne vehicle, the landing surface comprising: A system comprising a plurality of light sources arranged in a predetermined pattern, wherein a characteristic of the light emitted from each of said light sources is configured to be modulated as a function of time. Clause 2. The system of clause 1, wherein the light sources include a first set of light sources arranged in a first predetermined pattern, each of the light sources in the first set configured to be within a field of view of a camera associated with the aerial vehicle when the aerial vehicle is a first distance from the landing surface. Clause 3. The system of clause 2, wherein the light sources include a second set of light sources arranged in a second predetermined pattern, each of the light sources in the second set configured to be within the field of view of the camera when the aerial vehicle is a second distance from the landing surface. Clause 4. The system of clause 3, wherein each of the first set of light sources is configured to be outside the field of view of the camera when the aerial vehicle is at the second distance from the landing surface. Clause 5. The system of clause 3 or 4, wherein the area covered by the first set of light sources is larger than the area covered by the second set of light sources. Clause 6. A system according to any one of clauses 3 to 5, configured such that the intensity of the first set of light sources is greater than the intensity of the second set of light sources. Clause 7. A system as described in any one of clauses 1 to 6, wherein the predetermined pattern of light sources is associated with the landing surface. Clause 8. A system according to any one of clauses 1 to 7, wherein the light sources are arranged such that one of the light sources is uniquely identifiable in the predetermined pattern. Clause 9. A system as described in any one of clauses 1 to 8, wherein modulation of the characteristics of the emitted light is configured to identify the landing surface. Clause 10. A system according to any one of clauses 1 to 8, wherein the modulation of the characteristic of the emitted light is configured to identify one of the light sources. Clause 11. A system as described in any one of clauses 1 to 8, wherein the modulation of the characteristic of the emitted light is configured to identify the position of one of the light sources. Clause 12. A system as described in any one of clauses 1 to 8, wherein the modulation of the characteristic of the emitted light is configured to identify a condition of the landing surface. Clause 13. A system as described in any one of clauses 1 to 9, wherein the modulation of the characteristic of the emitted light is configured to encode a signal that authenticates the landing surface. Clause 14. A system according to any one of clauses 1 to 13, wherein the modulation of the properties of the emitted light comprises modulation of the intensity, frequency, amplitude, wavelength, phase, bandwidth or duty cycle of the emitted light. Clause 15. A system according to any one of clauses 1 to 14, wherein the wavelength of the light emitted from the light source is in the range of 800 nm to 850 nm. Clause 16. The system of clause 15, wherein the wavelength of the emitted light is about 810 nm. Clause 17. The system of clause 15, wherein the wavelength of the emitted light is about 1310 nm. Clause 18. The system of clause 15, wherein the wavelength of the emitted light is about 1550 nm. Clause 19. A system as described in any one of clauses 1 to 18, wherein the landing surface is a portable landing surface comprising a redeployable landing mat, cloth, or tarp. Clause 20. The system of any one of clauses 1-19, further comprising a controller circuit configured to operate the light source. Clause 21. A system according to any one of clauses 1 to 20, wherein each of the light sources is recessed relative to the landing surface. Clause 22. A system according to any one of clauses 1 to 21, wherein each of the light sources includes an optical sensor configured to detect a portion of light emitted from at least one other of the light sources. Clause 23. A system according to any one of clauses 1 to 22, further comprising a plurality of landing surfaces, each of the landing surfaces comprising a plurality of light sources arranged in a predetermined pattern, the characteristics of the light emitted from each of the light sources being configured to be modulated as a function of time. Clause 24. The system according to clause 23, wherein the landing surfaces are arranged horizontally offset from one another. Clause 25. The system of clause 23, wherein the landing surfaces are positioned vertically offset from one another. Clause 26. A system according to clause 23, wherein the landing surfaces are arranged horizontally and vertically offset from one another. Clause 27. A system comprising: An aerial vehicle, an airborne vehicle including a camera configured to generate an image based on information transmitted by a plurality of light sources adjacent a landing surface for the airborne vehicle; 1. A controller circuit comprising: receiving the generated image; a controller circuit configured to determine a position and orientation of the airborne vehicle based on the received image; The system wherein the light sources are arranged in a predetermined pattern on the landing surface, and the characteristics of the light emitted from each of the light sources are modulated with respect to time. Clause 28. The system of clause 27, wherein the camera is configured to provide a plan view of the light source on the landing surface. Clause 29. The system of clause 27, wherein the camera is configured to provide a forward view of the light source on the landing surface. Clause 30. A system described in any one of clauses 27 to 29, wherein the camera includes an optical filter, the optical filter configured to permit the wavelength range of the light emitted from each of the light sources. Clause 31. The system of clause 30, wherein the permitted wavelength range is between 800 nm and 850 nm. Clause 32. The system of clause 30, wherein the permitted wavelength range is about 810 nm. Clause 33. The system of clause 30, wherein the permitted wavelength range is about 1310 nm. Clause 34. The system of clause 30, wherein the permitted wavelength range is about 1550 nm. Clause 35. A system described in any one of clauses 30 to 34, wherein the optical filter includes a bandpass filter, the bandpass filter configured to permit the wavelength range of the light emitted from each of the light sources. Clause 36. A system according to any one of clauses 27 to 35, wherein the controller is further configured to adjust the capture rate of the camera based on the modulation rate of the light source. Clause 37. The system of clause 36, wherein adjusting the capture rate includes synchronizing the capture rate of the camera with the modulation rate of the light source. Clause 38. A system according to clause 36 or 37, wherein the capture rate of the camera is at least 100 frames per second (Hz). Clause 39. The system of any one of clauses 36 to 38, wherein the controller is further configured to adjust a flashing rate of the light source based on the capture rate. Clause 40. The system of clause 39, wherein the flashing rate of the light source is 30 Hz. Clause 41. The system of any one of clauses 36 to 40, wherein the controller is further configured to adjust the bit rate of the camera. Clause 42. A system according to clause 41, wherein the bit rate of the camera is 10 Hz or higher. Clause 43. A system according to any one of clauses 36 to 42, wherein the controller is further configured to send a synchronization pulse to synchronize the capture rate of the camera with the modulation rate of the light source. Clause 44. A system described in any one of clauses 27 to 43, wherein the modulation of the properties of the light emitted from each of the light sources comprises modulation of the intensity, frequency, amplitude, wavelength, phase, bandwidth, or duty cycle of the emitted light. Clause 45. A system as described in any one of clauses 27 to 44, wherein the camera is configured to be activated based on an activation signal from an external processor associated with the aerial vehicle, an operator of the aerial vehicle, or the controller. Clause 46. The system of any one of clauses 27 to 45, wherein the controller is further configured to generate an output signal, the output signal including information related to the position and the orientation of the airborne vehicle. Clause 47. The system of clause 46, wherein the information relating to the location of the airborne vehicle includes Global Positioning System (GPS) coordinates of the airborne vehicle. Clause 48. The system of clause 46 or 47, wherein the information relating to the orientation of the airborne vehicle includes an orientation of the airborne vehicle relative to the landing surface. Clause 49. The system of any one of clauses 46 to 48, wherein the controller is further configured to transmit the information relating to the position and the orientation of the airborne vehicle to the external processor. Clause 50. A system according to any one of clauses 27 to 49, wherein the camera is a colour, monochrome or hyperspectral camera. Clause 51. A system as described in any one of clauses 27 to 50, wherein the camera is configured to be activated after the airborne vehicle comes within a predetermined distance of the landing surface. Clause 52. A system according to any one of clauses 27 to 50, wherein the predetermined distance is 500 m or less. Article 53. A system comprising: 1. A system comprising: a plurality of light sources disposed on a landing surface of an airborne vehicle, wherein the arrangement of the light sources defines a set of intersecting imaginary lines, the light sources being disposed on each imaginary line, and wherein the distance between adjacent light sources on each imaginary line is unequal. Clause 54. The system according to clause 53, wherein an equal number of light sources are arranged on each imaginary line. Clause 55. A system according to clause 53 or 54, wherein the linear intersection ratio for each virtual line is independent of the viewing angle. Clause 56. A system according to any one of clauses 53 to 55, wherein the intersecting virtual lines define a plurality of regions, and the area intersection ratio for each region does not depend on the viewing angle. Clause 57. A method for estimating a pose for an airborne vehicle, comprising: providing a landing surface including light sources arranged in a predetermined pattern; modulating a characteristic of light emitted from said light source with respect to time; receiving an input signal related to the light emitted from the light source using a camera mounted on the airborne vehicle; generating an image of the light source based on the received input signal; determining a position and orientation of the airborne vehicle based on the image, wherein determining the position and orientation of the airborne vehicle comprises: Detecting at least one of the light sources in the image; determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is. Clause 58. The method of clause 57, further comprising encoding information relating to said at least one of the light sources into said characteristics of the modulated light emitted from said light sources. Clause 59. The method of clause 57 or 58, wherein determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is comprises using a processor to decode the encoded information relating to the light source. Clause 60. The method of any one of clauses 57 to 59, wherein detecting said at least one light source in said image further comprises background subtraction and thresholding. Clause 61. The method of any one of clauses 57 to 60, wherein detecting said at least one light source in said image further comprises image filtering techniques. Clause 62. The method of clause 61, wherein the image filtering comprises temporal filtering, spatial filtering, or a combination thereof. Clause 63. The method of any one of clauses 57 to 62, further comprising storing information relating to the predetermined pattern in a database. Clause 64. The method of clause 63, wherein determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is comprises matching points in the image with points in the database. Clause 65. A method according to any one of clauses 57 to 64, wherein determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is further comprises calculating a cross ratio of the positions of the corresponding light sources. Clause 66. A method according to any one of clauses 57 to 65, wherein the predetermined pattern includes an arrangement of the light sources defining a set of intersecting imaginary lines, the light sources being positioned on each imaginary line, and the distance between adjacent light sources on each imaginary line being unequal. Clause 67. The method of clause 66, wherein the linear intersection ratio of each imaginary line is independent of the viewing angle. Clause 68. A method according to clause 66 or 67, wherein the intersecting imaginary lines define a plurality of regions, and the area intersection ratio for each region is independent of the viewing angle. Clause 69. A method according to any one of clauses 57 to 68, wherein modulating the properties of the emitted light comprises modulating the intensity, frequency, amplitude, wavelength, phase, bandwidth or duty cycle of the emitted light. Clause 70. The method of any one of clauses 57 to 69, wherein determining the position and the orientation of the airborne vehicle comprises: Detecting at least four light sources in the image; determining which of the at least four light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the four light sources arranged in the predetermined pattern the detected light source is. Clause 71. The method of any one of clauses 57 to 69, wherein determining the position and the orientation of the airborne vehicle comprises: Detecting at least five of the light sources in the image; determining which of the at least five light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the five light sources arranged in the predetermined pattern the detected light source is. Clause 72. A computer-implemented system for estimating the pose of an airborne vehicle, comprising: a landing surface including light sources arranged in a predetermined pattern; at least one processor, wherein the at least one processor: modulating a characteristic of light emitted from said light source with respect to time; activating a camera mounted on the airborne vehicle to receive an input signal related to the light emitted from the light source; enabling the camera to generate an image of the light source based on the received input signal; determining a position and orientation of the airborne vehicle based on the generated image, wherein determining the position and orientation includes: Detecting at least one of the light sources in the image; determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is; determining the position and the orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is. Clause 73. A computer-implemented method for estimating the pose of an airborne vehicle, said method comprising the following operations performed by at least one processor: modulating, with respect to time, a characteristic of light emitted from light sources arranged in a predetermined pattern on a landing surface for the airborne vehicle; activating a camera mounted on the airborne vehicle to receive an input signal related to the light emitted from the light source; enabling the camera to generate an image of the light source based on the received input signal; determining a position and orientation of the airborne vehicle based on the image, wherein determining the position and orientation includes: Detecting at least one of the light sources in the image; determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is; determining the position and the orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is. Clause 74. The computer-implemented method of clause 73, wherein determining the position and the orientation of the airborne vehicle comprises: Detecting at least four light sources in the image; determining which of the at least four light sources arranged in the predetermined pattern the detected light source is; determining the position and the orientation of the airborne vehicle based on the determination of which of the four light sources arranged in the predetermined pattern the detected light source is. Clause 75. The computer-implemented method of clause 73 or 74, wherein determining the position and the orientation of the airborne vehicle comprises: Detecting at least five of the light sources in the image; determining which of the at least five light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the five light sources arranged in the predetermined pattern the detected light source is. Clause 76. A non-transitory computer-readable medium storing a set of instructions executable by at least one processor of an apparatus to cause said apparatus to perform a method, said method comprising: modulating, with respect to time, a characteristic of light emitted from light sources arranged in a predetermined pattern on a landing surface for the airborne vehicle; activating a camera mounted on the airborne vehicle to receive an input signal related to the light emitted from the light source; enabling the camera to generate an image of the light source based on the received input signal; determining a position and orientation of the airborne vehicle based on the image, wherein determining the position and orientation includes: Detecting at least one of the light sources in the image; determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is. Clause 77. The non-transitory computer-readable medium of clause 76, wherein the set of instructions executable by the at least one processor of the device causes the device to determine the position and the orientation of the airborne vehicle, and determining the position and the orientation of the airborne vehicle includes: Detecting at least four light sources in the image; determining which of the at least four light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the four light sources arranged in the predetermined pattern the detected light source is. Clause 78. The non-transitory computer-readable medium of clause 76, wherein the set of instructions executable by the at least one processor of the device causes the device to determine the position and the orientation of the airborne vehicle, and determining the position and the orientation of the airborne vehicle includes: Detecting at least five of the light sources in the image; determining which of the at least five light sources arranged in the predetermined pattern the detected light source is; and determining the position and the orientation of the airborne vehicle based on the determination of which of the five light sources arranged in the predetermined pattern the detected light source is. Clause 79. A computer-implemented method for locating light sources on a landing surface for an airborne vehicle, said method comprising the following operations executed by at least one processor, said operations comprising: activating a camera mounted on the airborne vehicle to receive input signals related to light emitted from the light sources arranged in a predetermined pattern on the landing surface for the airborne vehicle, the light having a time-modulated characteristic; enabling the camera to generate at least two images of the light source based on the received input signal; enabling a detector to detect at least one of the light sources in the at least two images using a detection algorithm, the detection algorithm comprising: determining the difference in pixel intensity values ​​of said at least two images using a subtraction algorithm; applying a predetermined threshold pixel intensity to the difference frame to generate a mask; using the mask to identify the location of pixels in the image that represent the light source; and calculating a position of the light source on the landing surface based on the position of the pixel in the image using a centroid algorithm. Clause 80. The computer-implemented method of clause 79, wherein the detection algorithm further comprises generating the difference frame based on the determined difference in the pixel intensity values ​​of the at least two images. Clause 81. The computer-implemented method of clause 79 or 80, wherein the characteristics of light are adjusted such that the light source for a first image of the at least two images is fully activated and the light source for a second image of the at least two images is fully deactivated. Clause 82. The computer-implemented method of clause 81, wherein the first image includes a signal image and the second image includes a background image. Clause 83. The computer-implemented method of clause 81 or 82, wherein the first image and the second image comprise consecutive images. Clause 84. The computer-implemented method of any one of clauses 81 to 83, wherein the detection algorithm further comprises performing image registration to align the first image and the second image to enable background subtraction. Clause 85. The computer-implemented method of clause 84, wherein the image registration is performed using techniques including feature matching, transformation matching, current state estimation, or a combination thereof. Clause 86. The computer-implemented method of any one of clauses 79 to 85, wherein the detection algorithm further comprises tracking the identified location of the pixel representing the light source by extrapolating based on a combination of speed information of the airborne vehicle and the elapsed time between capturing the at least two images. Clause 87. The computer-implemented method of any one of clauses 79-86, wherein the airborne vehicle comprises an electric vertical take-off and landing aircraft. Clause 88. A computer-implemented method for mapping locations in an image to locations on a landing surface for an airborne vehicle, said method comprising the following operations performed by at least one processor: using a detection algorithm to detect a light source in the image, the light source being disposed on the landing surface and configured to emit light detectable by a camera mounted on the airborne vehicle; using a first association algorithm to associate locations in the image representing the detected light sources with corresponding locations of the light sources on the landing surface, the association algorithm comprising: normalizing the locations within the image representing the detected light sources in Cartesian coordinate space; transforming the normalized positions into curves in polar coordinate space, wherein collinear normalized positions in the Cartesian coordinate space form curves that intersect at a common point in the polar coordinate space; discretizing the polar coordinate space into a plurality of bins, each bin being represented by a value indicating the number of times a curve passes through the bin's location; upon determining whether the bin value exceeds a predetermined threshold, transforming the location of the bin in the polar coordinate space to the Cartesian coordinate space; forming lines in the Cartesian coordinate space, each line connecting at least a plurality of points equal to the value of the corresponding bin; using a clustering algorithm to group substantially parallel lines and form a rectangular frame in integer grid space from the grouped lines; computing a homography matrix configured to move the points from the Cartesian coordinate space to the integer grid; and using the computed homography matrix to map each point onto the integer grid. Clause 89. The computer-implemented method of clause 88, wherein normalizing the positions in the image includes constructing a transformation matrix to calculate the mean of the positions and setting the variance of the positions to one. Clause 90. The computer-implemented method of clause 88 or 89, wherein normalizing the position within the image further comprises rotating the Cartesian coordinate space by an angle to compensate for rotation caused by an approach angle of the airborne vehicle toward the landing surface. Clause 91. The computer-implemented method of any one of clauses 88 to 90, wherein the value of the bin is increased by one for each instance of the curve that passes through the position of the bin. Clause 92. The computer-implemented method of any one of clauses 88-91, wherein the first association algorithm further comprises refining one or more lines by rejecting one or more lines in the Cartesian coordinate space based on a fit to the detected position of the light source in the image. Clause 93. The computer-implemented method of clause 92, wherein the first association algorithm further comprises iteratively refining the one or more lines. Clause 94. The computer-implemented method of any one of clauses 88 to 93, further comprising labeling each location on the integer grid with a reference character. Clause 95. The computer-implemented method of clause 94, wherein the labeling is based on a predefined sequence. Clause 96. The computer-implemented method of any one of clauses 88 to 95, wherein mapping each point onto the integer grid indicates an offset distance, the offset distance being the distance between a reference position on the integer grid and a corresponding mapped point. Clause 97. The computer-implemented method of clause 96, further comprising rejecting false detections from the association based on the offset distance. Clause 98. The computer-implemented method of clause 97, wherein rejecting the false detection includes comparing the offset distance to a threshold offset distance. Clause 99. The computer-implemented method of clause 98, further comprising: identifying the mapped points where the offset distance is greater than the threshold offset distance as false detections; The computer-implemented method further includes rejecting the association upon determining that the number of false detections exceeds a tolerance threshold. Clause 100. The computer-implemented method of any one of clauses 88 to 99, further comprising using a second association algorithm to associate locations in the image representing the detected light sources with the corresponding locations of the light sources on the landing surface. Clause 101. The computer-implemented method of clause 100, wherein the first association algorithm comprises a grid association algorithm and the second association algorithm comprises an iterative closest point (ICP) algorithm, a thin plate spline robust point matching (TPS-RPM) algorithm, a point tracking algorithm, or a combination thereof. Article 102. A system comprising: a first plurality of light sources arranged in a predetermined pattern on a landing surface for the airborne vehicle; a second plurality of light sources positioned along a flight path of the airborne vehicle to guide the airborne vehicle, wherein a characteristic of light emitted from each of the first and second plurality of light sources is configured to be modulated with respect to time, and wherein the modulated light emitted from each of the first and second plurality of light sources is configured to be detectable by a camera mounted on the airborne vehicle. Clause 103. The system of clause 102, wherein the second plurality of light sources are located on a building, pole, tower, natural structure, or roof of a structure along the flight path. Article 104. A system comprising: a landing surface for an airborne vehicle, the landing surface comprising: 1. A system including a predetermined pattern of linear light sources on the landing surface for the airborne vehicle, wherein a characteristic of light emitted from the plurality of linear light sources is configured to be modulated with respect to time, the modulation configured to encode a signal representative of the identity of the landing surface. Clause 105. The system of clause 104, further comprising a plurality of point light sources arranged in said predetermined pattern on said landing surface for said airborne vehicle, wherein the system is configured such that the characteristics of light emitted from said plurality of point light sources are modulated with respect to time. Clause 106. A system according to clause 104 or 105, wherein the shape of the predetermined pattern of linear light sources is rectangular. Clause 107. The system of clause 106, wherein an edge of the rectangular predetermined pattern of the linear light source comprises two or more collinear line segments, each line segment configured to represent a single bit of information based on the activation state of the linear light source within the line segment. Clause 108. The system of any one of clauses 104 to 107, further comprising an aerial vehicle including a camera mounted on the aerial vehicle, the camera configured to receive the encoded signal. Clause 109. The system of clause 108, further comprising a processor communicatively associated with the camera, the processor configured to decode the encoded signal received by the camera and generate an output based on the decoding. Clause 110. A system according to any one of clauses 106 to 109, wherein the shape of the predetermined pattern of the linear light source is a triangle, a circle, an ellipse, a polygon, or a combination thereof. Article 111. A method for identifying a landing surface for an airborne vehicle, said method comprising: receiving, by a camera mounted on the airborne vehicle, encoded signals from a plurality of linear light sources arranged in a first predetermined pattern on the landing surface for the airborne vehicle, the encoded signals representing an identity of the landing surface, the encoded signals being encoded by modulating a characteristic of light emitted from the linear light sources; and decoding the received encoded signal using a processor associated with the camera to generate an output including information related to the identity of the landing surface; The method of claim 1, wherein the first predetermined pattern of linear light sources comprises collinear line segments, each line segment configured to represent a single bit of information based on an activation state of the linear light sources within the line segment. Clause 112. The method of clause 111, wherein in a first activation state, the line segment represents a bit value of 1, and in a second activation state, the line segment represents a bit value of 0. Clause 113. The method of clause 112, wherein the first activation state is an ON state and the second activation state is an OFF state of the linear light source within the line segment. Clause 114. A method according to any one of clauses 111 to 113, further comprising receiving, by the camera mounted on the aerial vehicle, light emitted from a point light source in a second predetermined pattern on the landing surface for the aerial vehicle, wherein the characteristics of the light emitted from the point light source are modulated with respect to time. Clause 115. A method for estimating the pose of an airborne vehicle, said method comprising: providing a constellation of light sources on a portable landing surface at a landing site for said airborne vehicle; calibrating the relative positions of the light sources using ultra-wideband signals between the light sources to determine the configuration of the constellation; transmitting information related to the determined constellation configuration of light sources to an airborne vehicle approaching the landing site. Clause 116. The method of clause 115, further comprising estimating the pose of the aerial vehicle based on the constellation configuration and an image of the landing surface captured by a camera mounted on and associated with the aerial vehicle. Clause 117. The method of clause 115 or 116, wherein the portable landing surface comprises a rapidly deployable landing surface, a redeployable landing surface, a rollable mat, a cloth, a tarp, a net, or a mesh. Clause 118. A method according to any one of clauses 115 to 117, wherein the portable landing surface is configured to match the contours of the landing site. Clause 119. The method of any one of clauses 115 to 118, wherein the constellation of light sources includes a point light source and a linear light source. Clause 120. The method of any one of clauses 115 to 119, wherein said constellation of light sources includes remotely operable battery-powered light sources. Clause 121. A method for estimating the pose of an airborne vehicle, said method comprising: providing a constellation of light sources on a portable landing surface in a predetermined pattern; placing the portable landing surface at a contoured landing site, the portable landing surface configured to fit the contoured landing site; estimating a configuration of a constellation of light sources on the deployed portable landing surface; and estimating the pose of the airborne vehicle based on the estimated constellation configuration and an image of the landing surface captured by a camera mounted on and associated with the airborne vehicle. Clause 122. The method of clause 121, wherein the portable landing surface comprises a rapidly deployable landing surface, a redeployable landing surface, a rollable mat, a cloth, a tarp, a net, or a mesh. Clause 123. The method of clause 121 or 122, wherein said constellation of light sources includes a combination of point and linear light sources. Clause 124. The method of any one of clauses 120 to 123, wherein the constellation of light sources includes a point light source and a linear light source. Clause 125. The method of any one of clauses 121 to 124, wherein said constellation of light sources includes remotely operable battery-powered light sources. Article 126. Air vehicles, a camera configured to generate an image based on information received from a plurality of light sources located on a landing surface for the airborne vehicle; a processor associated with the camera and configured to receive the image and perform the following operations: Detecting a light source in the image using a detection algorithm, the light source being disposed on the landing surface and configured to emit light detectable by the camera; performing an association between locations in the image representing the detected light sources and corresponding locations of the light sources on the landing surface, the processor being configured to perform the association in a first mode of operation and a second mode of operation; executing one or more association algorithms in the first mode of operation to generate a confidence score for the association; executing one or more tracking algorithms in the second mode of operation based on the confidence scores obtained from the first mode of operation; and determining a position or orientation of the airborne vehicle based on the performed association. Clause 127. The airborne vehicle of clause 126, wherein the processor is configured to automatically switch between the first operational mode and the second operational mode based on a predetermined threshold confidence score. Clause 128. The airborne vehicle of clause 127, wherein the processor is configured to request user input to switch between the first operational mode and the second operational mode based on the predetermined threshold confidence score. Clause 129. An airborne vehicle as described in any one of clauses 126 to 128, wherein the processor is configured to sequentially execute the first operating mode and the second operating mode. Clause 130. An airborne vehicle according to any one of clauses 126 to 129, wherein the processor: switching from the first operating mode to the second operating mode; and after switching from the first operational mode to the second operational mode, executing the first operational mode and the second operational mode in parallel. Clause 131. The airborne vehicle of any one of clauses 126 to 130, wherein the one or more association algorithms include a grid association algorithm, a thin-plate spline robust point matching (TPS-RPM) association algorithm, or an iterative closest point (ICP) algorithm. Clause 132. The airborne vehicle of any one of clauses 126 to 131, wherein the one or more tracking algorithms include local association point tracking or pose-based point tracking. Clause 133. An airborne vehicle according to any one of clauses 126 to 132, wherein the detection algorithm is configured to detect modulation of a characteristic of the plurality of light sources over time. Clause 134. The airborne vehicle of clause 133, wherein the plurality of light sources includes a combination of linear light sources and point light sources. Clause 135. An airborne vehicle according to clause 134, wherein the linear light source and the point light source are arranged in the predetermined pattern on the landing surface. Clause 136. The airborne vehicle of clause 135, wherein the landing surface comprises a portable landing surface, a rollable landing surface, a redeployable landing surface, a tarp, a net, a mesh, or a combination thereof. Clause 137. An aerial vehicle according to any one of clauses 126 to 136, wherein executing the one or more association algorithms comprises: normalizing the locations within the image representing the detected light sources in Cartesian coordinate space; transforming the normalized positions into curves in polar coordinate space, wherein collinear normalized positions in the Cartesian coordinate space form curves that intersect at a common point in the polar coordinate space; discretizing the polar coordinate space into a number of bins, each bin being represented by a value indicating the number of times a curve passes through the bin's location; upon determining whether the bin value exceeds a predetermined threshold, transforming the location of the bin in the polar coordinate space to the Cartesian coordinate space; forming lines in the Cartesian coordinate space, each line connecting at least a plurality of points equal to the value of the corresponding bin; using a clustering algorithm to group substantially parallel lines and form a rectangular frame in integer grid space from the grouped lines; computing a homography matrix configured to move the points from the Cartesian coordinate space to the integer grid; and using the calculated homography matrix to map each point onto the integer grid. Clause 138. The airborne vehicle of clause 137, wherein the processor is further configured to normalize the positions in the image by constructing a transformation matrix to calculate the mean of the positions and setting the variance of the positions to one. Clause 139. The airborne vehicle of clause 137, wherein the processor is further configured to normalize the position within the image by rotating the Cartesian coordinate space by an angle to compensate for rotation caused by an approach angle of the airborne vehicle toward the landing surface. Clause 140. The airborne vehicle of any one of clauses 137 to 139, wherein the processor is further configured to increment the value of the bin by one for each instance of a curve passing through the position of the bin. Clause 141. An aerial vehicle as described in any one of clauses 137 to 140, wherein the processor is further configured to refine one or more lines by rejecting one or more lines in the Cartesian coordinate space based on a fit to the detected position of the light source in the image. Clause 142. An airborne vehicle according to any one of clauses 137 to 141, wherein the processor is further configured to label each location on the integer grid with a reference character, the label being based on a predefined sequence. Clause 143. An airborne vehicle according to any one of clauses 137 to 142, wherein the processor is further configured to map each point onto the integer grid indicating an offset distance, the offset distance being the distance between a reference position on the integer grid and a corresponding mapped point. Clause 144. An airborne vehicle as described in any one of clauses 137 to 143, wherein the processor is further configured to reject false detections from the association based on the offset distance, and wherein the rejection of the false detections includes comparing the offset distance to a threshold offset distance. Clause 145. The airborne vehicle of any one of clauses 137 to 144, wherein the processor is further configured to identify the mapped points where the offset distance is greater than the threshold offset distance as false detections, and to reject the association upon determining that the number of false detections exceeds an acceptable threshold. Clause 146. An airborne vehicle according to any one of clauses 126 to 145, wherein determining the position of the airborne vehicle or the orientation of the airborne vehicle is further based on information from one or more of a Global Positioning System (GPS) or an Inertial Navigation System (INS). Clause 147. The airborne vehicle of any one of clauses 126 to 146, wherein determining one of the position and orientation of the airborne vehicle based on the performed association includes determining both the position and the orientation of the airborne vehicle. Clause 148. An airborne vehicle according to any one of clauses 126 to 147, Lift propellers and a controller configured to actuate the lift propeller based on the determined position or orientation of the airborne vehicle. Clause 149. An airborne vehicle according to any one of clauses 126 to 148, Tilt propellers and a controller configured to actuate the tilt propeller based on the determined position or orientation of the airborne vehicle. Clause 150. An airborne vehicle as set forth in any one of clauses 126 to 149, A tilt actuator, a controller configured to actuate the tilt actuator based on the determined position or orientation of the airborne vehicle. Clause 151. An airborne vehicle according to any one of clauses 126 to 150, A control surface; a controller configured to actuate the control surface based on the determined position or orientation of the airborne vehicle. Clause 152. The airborne vehicle of clause 151, wherein the control surface comprises one of a flaperon and a ruddervator. Article 153. A navigation system for an airborne vehicle, comprising: a camera configured to generate an image based on information received from a plurality of light sources arranged in a predetermined pattern on a landing surface for the airborne vehicle; a processor associated with the camera and configured to receive the image and perform the following operations: activating, using the processor, a camera mounted on the airborne vehicle to receive input signals associated with light emitted from the light sources arranged in a predetermined pattern on the landing surface for the airborne vehicle, the light having a time-modulated characteristic; enabling the camera to generate at least two images of the light source based on the received input signal; detecting the light source in the at least two images using a detection algorithm; performing an association between locations in the image representing the detected light sources and corresponding locations of the light sources on the landing surface, the processor being configured to perform the association in a first mode of operation and a second mode of operation; executing one or more association algorithms in said first mode of operation; executing one or more tracking algorithms in the second mode of operation based on results obtained from the first mode of operation; and determining one of a position and an orientation of the airborne vehicle based on the performed association. Clause 154. A navigation system according to clause 153, comprising: The navigation system further includes the plurality of light sources arranged in the predetermined pattern on the landing surface. Clause 155. A navigation system according to clause 154, wherein the characteristics of the light emitted from each of said light sources are modulated with respect to time. Clause 156. The navigation system of any one of clauses 153 to 155, further comprising a controller configured to actuate a lift propeller based on the determined position or heading of the airborne vehicle. Clause 157. The navigation system of any one of clauses 153 to 156, further comprising a controller configured to actuate a tilt propeller based on the determined position or heading of the airborne vehicle. Clause 158. The navigation system of any one of clauses 153 to 157, further comprising a controller configured to actuate a tilt actuator based on the determined position or heading of the airborne vehicle. Clause 159. The navigation system of any one of clauses 153 to 158, further comprising a controller configured to actuate a control surface based on the determined position or heading of the airborne vehicle. Clause 160. The navigation system of clause 159, wherein the control surface includes one of a flaperon and a ruddervator. Article 161. A system, a landing surface for an airborne vehicle; a plurality of light sources arranged in a predetermined pattern, wherein a characteristic of the light emitted from each of the light sources is configured to be modulated with respect to time; the plurality of light sources include linear light sources and point light sources, The system, wherein the landing surface comprises a portable landing surface. Clause 162. The system of clause 161, further comprising: a first processor configured to modulate the characteristics of the light emitted from the light source with respect to time; a second processor, activating a camera mounted on the aerial vehicle to receive an input signal related to the light emitted from the light source; enabling the camera to generate an image of the light source based on the received input signal; and determining one of a position and an orientation of the airborne vehicle based on the generated image, wherein determining the position or the orientation includes: Detecting at least one of the light sources in the image; determining which of the at least one of the light sources arranged in the predetermined pattern the detected light source is; and determining the position or the orientation of the airborne vehicle based on the determination of which of the at least one of the light sources arranged in the predetermined pattern the detected light source is. Clause 163. The system of clause 162, further comprising a controller configured to actuate a lift propeller based on the determined position or orientation of the airborne vehicle. Clause 164. The system of clause 162 or 163, further comprising a controller configured to actuate a tilt propeller based on the determined position or orientation of the airborne vehicle. Clause 165. The system of any one of clauses 162 to 164, further comprising a controller configured to actuate a tilt actuator based on the determined position or orientation of the airborne vehicle. Clause 166. The system of any one of clauses 162 to 165, further comprising a controller configured to actuate a control surface based on the determined position or orientation of the airborne vehicle. Clause 167. The system of any one of clauses 162 to 166, wherein the control surface includes one of a flaperon and a ruddervator. Clause 168. A system according to any one of clauses 161 to 167, wherein the portable landing surface is configured to match the contours of the landing site. Clause 169. A system according to any one of clauses 161 to 168, wherein the plurality of light sources includes a remotely operable battery-powered light source. Clause 170. The system of any one of clauses 161 to 169, wherein the portable landing surface comprises a redeployable landing surface, a rollable mat, a cloth, a tarp, a net, or a mesh. Clause 171. A system described in any one of clauses 161 to 169, wherein the second processor is further configured to calibrate the relative positions of the light sources using the ultra-wideband signals between the light sources. Clause 172. A system according to any one of clauses 161 to 171, wherein the plurality of light sources includes a plurality of infrared light sources. Clause 173. A computer-readable medium storing instructions which, when executed by at least one processor of a device, cause said device to perform the method of any one of clauses 57-69, 71-75, 79-101, 111-125.

[0326] The foregoing description has been presented for purposes of illustration. It is not exhaustive and is not intended to limit the invention to the precise forms or embodiments disclosed. Modifications and adaptations of the invention will be apparent to those skilled in the art from consideration of the specification and practice of embodiments of the invention disclosed herein.

Claims

1. An aerial vehicle, a camera configured to emit light detectable by the camera and configured to generate an image based on information received from a plurality of light sources disposed on a landing surface for the airborne vehicle; a processor associated with the camera and configured to receive the image and perform the following operations: detecting a light source in the image using a detection algorithm; performing an association between locations in the image representing the detected light sources and corresponding locations of the light sources on the landing surface, the processor being configured to perform the association in a first mode of operation and a second mode of operation; running one or more association algorithms in the first mode of operation to generate a confidence score for the association; executing one or more tracking algorithms in the second mode of operation based on the confidence scores obtained from the first mode of operation; and determining one of a position and an orientation of the airborne vehicle based on the performed association.

2. 10. The airborne vehicle of claim 1, wherein the processor is configured to automatically switch between the first and second modes of operation based on a predetermined threshold confidence score.

3. 3. The airborne vehicle of claim 2, wherein the processor is configured to request user input to switch between the first operational mode and the second operational mode based on the predetermined threshold confidence score.

4. 4. The airborne vehicle of claim 1, wherein the processor is configured to sequentially execute the first and second operational modes.

5. 5. The airborne vehicle of claim 1, wherein the processor: switching from the first mode of operation to the second mode of operation; and after switching from the first operational mode to the second operational mode, executing the first operational mode and the second operational mode in parallel.

6. 6. The airborne vehicle of claim 1, wherein the one or more association algorithms comprise a grid association algorithm, a thin-plate spline robust point matching (TPS-RPM) association algorithm, or an iterative closest point (ICP) algorithm.

7. The airborne vehicle of any one of claims 1 to 6, wherein the one or more tracking algorithms include local association point tracking or pose-based point tracking.

8. 8. An airborne vehicle according to any one of claims 1 to 7, wherein the detection algorithm is configured to detect modulation of a characteristic of the plurality of light sources over time.

9. The airborne vehicle of claim 8 , wherein the plurality of light sources includes a combination of linear and point light sources.

10. 9. The airborne vehicle of claim 1, wherein executing the one or more association algorithms comprises: normalizing the locations within the image representing the detected light sources in Cartesian coordinate space; transforming the normalized positions into curves in polar coordinate space, wherein collinear normalized positions in the Cartesian coordinate space form curves that intersect at a common point in the polar coordinate space; discretizing the polar coordinate space into a number of bins, each bin being represented by a value indicating the number of times a curve passes through the bin's location; upon determining whether the bin value exceeds a predetermined threshold, transforming the location of the bin in the polar coordinate space to the Cartesian coordinate space; forming lines in the Cartesian coordinate space, each line connecting at least a plurality of points equal to the value of the corresponding bin; using a clustering algorithm to group substantially parallel lines and form a rectangular frame for integer grid space from the grouped lines; computing a homography matrix configured to move the points from the Cartesian coordinate space to the integer grid; and using the calculated homography matrix to map each point onto the integer grid.

11. 11. The airborne vehicle of claim 10, wherein the processor is configured to normalize the positions in the image by constructing a transformation matrix to calculate a mean of the positions and setting a variance of the positions to one.

12. 12. The airborne vehicle of claim 10 or 11, wherein the processor is configured to normalize the position within the image by rotating the Cartesian coordinate space by an angle to compensate for rotation caused by an approach angle of the airborne vehicle toward the landing surface.

13. 13. An airborne vehicle according to any one of claims 10 to 12, wherein the processor is configured to increment the value of the bin by one for each instance of a curve passing through the position of the bin.

14. 14. The airborne vehicle of claim 10, wherein the processor is configured to refine one or more lines by rejecting one or more lines in the Cartesian coordinate space based on a fit to the detected positions of the light sources in the image.

15. 15. The airborne vehicle of any one of claims 10 to 14, wherein the processor is configured to label each location on the integer grid with a reference letter, the labels being based on a predefined sequence.

16. 16. The airborne vehicle of any one of claims 10 to 15, wherein the processor is configured to map each point onto the integer grid indicating an offset distance, the offset distance being the distance between a reference position on the integer grid and a corresponding mapped point.

17. 17. The airborne vehicle of claim 10, wherein the processor is configured to reject false detections from the association based on the offset distance, and wherein the rejection of the false detections comprises comparing the offset distance to a threshold offset distance.

18. 18. The airborne vehicle of claim 10, wherein the processor is configured to identify the mapped points where the offset distance is greater than the threshold offset distance as false detections, and to reject the association upon determining that a number of false detections exceeds a tolerance threshold.

19. 19. The airborne vehicle of any one of claims 1 to 18, wherein determining the position of the airborne vehicle or the orientation of the airborne vehicle is further based on information from one or more of a Global Positioning System (GPS) or an Inertial Navigation System (INS).

20. An aerial vehicle according to any one of claims 1 to 19, a controller configured to actuate components of the airborne vehicle based on the determined position or orientation of the airborne vehicle; The airborne vehicle, wherein the component includes one of a lift propeller, a tilt propeller, a tilt actuator, and a control surface.

21. 1. A method of operating an aerial vehicle, comprising: generating an image with a camera based on information received from a plurality of light sources located on a landing surface for the airborne vehicle; Detecting a light source in the image using a detection algorithm, the light source being disposed on the landing surface and configured to emit light detectable by the camera; performing an association between locations in the image representing the detected light sources and corresponding locations of the light sources on the landing surface, the association comprising a first mode of operation and a second mode of operation; the first mode of operation includes executing one or more association algorithms to generate a confidence score for the association; the second mode of operation includes executing one or more tracking algorithms based on the confidence scores obtained from the first mode of operation; and determining a position or orientation of the airborne vehicle based on the performed association.

22. 1. A navigation system for an airborne vehicle, comprising: a camera configured to generate an image based on information received from a plurality of light sources arranged in a predetermined pattern on a landing surface for the airborne vehicle; a processor associated with the camera and configured to receive the image and perform the following operations: activating, using the processor, a camera mounted on the airborne vehicle to receive input signals associated with light emitted from the light sources arranged in a predetermined pattern on the landing surface for the airborne vehicle, the light having a time-modulated characteristic; enabling the camera to generate at least two images of the light source based on the received input signal; detecting the light source in the at least two images using a detection algorithm; performing an association between locations in the image representing the detected light sources and corresponding locations of the light sources on the landing surface, the processor being configured to perform the association in a first mode of operation and a second mode of operation; executing one or more association algorithms in the first mode of operation; executing one or more tracking algorithms in the second mode of operation based on results obtained from the first mode of operation; and determining one of a position and an orientation of the airborne vehicle based on the performed association.

23. 23. The navigation system of claim 22, The navigation system further includes the plurality of light sources arranged in the predetermined pattern on the landing surface.

24. 24. The navigation system of claim 23, wherein the landing surface comprises a vertiport landing surface.

25. 24. The navigation system of claim 23, wherein the landing surface comprises a portable landing surface comprising one of a redeployable landing surface, a rollable mat, a cloth, a tarp, a net, and a mesh.

26. A navigation system according to any one of claims 23 to 25, wherein the properties of the light emitted from each of the light sources are modulated with respect to time.

27. A navigation system according to any one of claims 22 to 26, a controller configured to actuate components of the airborne vehicle based on the determined position or orientation of the airborne vehicle; The navigation system, wherein the component includes one of a lift propeller, a tilt propeller, a tilt actuator, and a control surface.

28. 1. A system comprising: a portable landing surface for an airborne vehicle, said portable landing surface comprising: a plurality of light sources arranged in a predetermined pattern, wherein a characteristic of the light emitted from each of the light sources is configured to be modulated with respect to time; The system, wherein the plurality of light sources includes a linear light source and a point light source.

29. 30. The system of claim 28, comprising a processor configured to calibrate the relative positions of the light sources using ultra-wideband signals between the light sources.

30. 30. The system of claim 28 or 29, wherein the plurality of light sources comprises a plurality of infrared light sources.

Images