Event-assisted unmanned aerial vehicle night precise landing system and method

Abstract

The invention discloses an event-assisted unmanned aerial vehicle night precise landing system and method. The system comprises a high-frequency dynamic infrared sign array module, an event camera module, a pose resolving module and a landing control module. The system takes a high-frequency dynamic infrared mark array as a positioning reference, an event camera carried on the unmanned aerial vehicle captures an optical event flow with unique time sequence characteristics by accurately controlling the independent flicker frequency of each infrared light source, then the pose of the unmanned aerial vehicle is calculated by using an N-point perspective method, and finally accurate landing is realized by a landing control module. According to the invention, the event camera and the high-frequency dynamic infrared sign array are innovatively combined, the cooperative working mechanism of the event camera and the high-frequency dynamic infrared sign array is utilized, sub-millisecond-level positioning feedback is realized in a complex environment, the limitation of a traditional static optical sign is broken through, the positioning precision and reliability are improved, and the method is suitable for accurate take-off, landing and positioning of the unmanned aerial vehicle at night and in the complex environment.

Description

An event-assisted precision night landing system and method for unmanned aerial vehicles Technical Field

[0001] This invention relates to unmanned aerial vehicle (UAV) technology, and in particular to an event-assisted nighttime precision landing system and method for UAVs. Background Technology

[0002] As one of the fastest-growing core projects in the field of robotics, drones have become a focus of attention for many industries due to their flexibility and efficiency. Currently, drones are widely used in the low-altitude economy, undertaking diverse tasks such as logistics delivery, rapid rescue, and infrastructure inspection. However, as drone applications gradually expand into urban environments, their operation faces greater safety challenges. The dense building structures and busy pedestrian activity in urban environments place higher demands on the navigation and positioning accuracy of drones. To ensure the safety of urban buildings and pedestrians, the drone's position must be precisely controlled within a preset flight path to avoid the risks of deviation and collision. Simultaneously, in the case of multiple drones operating collaboratively, accurate positioning is also a crucial prerequisite for ensuring mission coordination and flight safety. Therefore, drone positioning technology plays a vital role in achieving safe and reliable low-altitude operations and is one of the core areas of current drone technology research and development.

[0003] In related technologies, UAV positioning primarily relies on the Global Navigation Satellite System (GNSS) for absolute positioning. Operating UAVs are typically equipped with signal receivers, which receive satellite signals and process the corresponding data to obtain the UAV's absolute position information. In addition, UAVs can also be equipped with sensors such as cameras, inertial measurement units, and barometers. During flight, onboard computing equipment processes the input data from these sensors and uses methods such as Simultaneous Localization and Mapping (SLAM) to estimate its own pose. This method enables UAVs to locate and map in unknown environments, improving their autonomy and reducing reliance on satellite signals. Furthermore, UAV positioning can also rely on pre-built high-resolution 3D maps, achieving positioning by matching LiDAR or visual data with the map.

[0004] While these existing technologies have improved the positioning capabilities of drones to some extent, they still have many limitations in complex urban environments. For example, GNSS signals are easily blocked and interfered with in urban canyons and areas with tall buildings, leading to a decrease in positioning accuracy; real-time positioning and mapping technologies may not be able to accurately map and locate in dynamic environments or scenarios with sparse feature points; and positioning methods based on high-definition 3D maps rely on real-time map updates and high-precision matching, making them difficult to adapt to rapidly changing urban environments.

[0005] Drone positioning technology plays a crucial role in ensuring the safety and reliability of low-altitude operations. However, existing positioning technologies, especially those relying on Global Navigation Satellite Systems (GNSS), are prone to signal instability in environments with tall buildings, such as urban canyons, due to signal blockage. Furthermore, when drones enter densely populated residential areas, they may encounter signal congestion due to excessive signal access from service providers, further affecting positioning accuracy. Moreover, these satellite navigation-based positioning solutions typically only provide meter-level accuracy, which is insufficient to meet the demands of high-precision positioning.

[0006] Meanwhile, while real-time positioning and mapping technologies based on airborne sensors can provide more accurate positioning information, they are vulnerable to errors caused by sensor noise. In large-scale outdoor scenarios, these technologies have high computational resource requirements, resulting in slow processing speeds, high resource consumption, and a tendency to accumulate errors and drift over extended periods. Although existing positioning methods attempt to improve accuracy by utilizing high-precision urban maps or point cloud maps in conjunction with airborne sensor input for registration, ordinary radar or visual sensors are prone to tracking loss issues with high-speed moving drones due to limitations in update frequency.

[0007] Traditional visual sensors face fundamental technical obstacles at night. Commercial RGB cameras experience a sharp drop in signal-to-noise ratio and a reduction in effective imaging distance of over 50% when illumination falls below 1 lux. Even with high-sensitivity starlight-level cameras, high-power supplemental lighting is still required to obtain usable images, increasing power consumption and posing significant exposure risks in military applications. While infrared thermal imaging cameras possess night vision capabilities, their resolution is generally low, making it difficult to meet centimeter-level positioning requirements. The discernibility of natural features decreases in nighttime environments. Real-world testing data shows that the feature matching success rate of ORB-SLAM2-based visual systems in nighttime scenes plummets from 85% during the day to 30%, directly leading to broken positioning trajectories.

[0008] It should be noted that the information disclosed in the background section above is only for understanding the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0009] The main objective of this invention is to overcome the deficiencies in the aforementioned background technology and provide an event-assisted precision night landing system and method for UAVs, offering a high-precision, high-real-time, and highly robust active guidance landing solution for UAVs in GNSS denial and low-light / no-light night environments.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, an event-assisted precise night landing system for unmanned aerial vehicles (UAVs) includes: a high-frequency dynamic infrared marker array module, composed of multiple independently controllable infrared light sources, which generates optical signals with unique temporal codes through a preset three-dimensional geometric layout; an event camera module, mounted on the UAV, used to capture the optical event stream of the infrared light sources and extract the frequency features of the light sources based on high temporal resolution; a pose calculation module, connected to the event camera module, used to calculate the UAV pose by combining the two-dimensional projection coordinates of the matching light source on the imaging plane of the event camera with known three-dimensional spatial coordinates using the N-point perspective method (PnP); and a landing control module, which controls the UAV trajectory based on the solved real-time pose of the UAV to achieve precise night landing; wherein, through the coordinated operation of the independent light source flashing frequency of the infrared marker array module and the microsecond-level response characteristics of the event camera module, sub-millisecond-level positioning feedback is achieved in complex environments.

[0011] In a second aspect of the invention, a method for precise nighttime landing of a UAV based on the system described herein includes the following steps: S1. Dynamic feature matching: capturing the optical event stream of an infrared marker array using an onboard event camera, and extracting the unique frequency features of each light source based on the time difference histogram analysis of the event stream to achieve light source identification and matching; S2. Real-time pose calculation: calculating the UAV pose by combining the two-dimensional projection coordinates of the matched light source on the imaging plane of the event camera with the known three-dimensional spatial coordinates using the N-point perspective method (PnP); S3. Landing control: controlling the UAV trajectory based on the solved real-time UAV pose to achieve precise nighttime landing.

[0012] This invention offers the following advantages: It proposes an event-assisted nighttime precision landing system and method for unmanned aerial vehicles (UAVs). Through a technical solution for precise UAV takeoff, landing, and positioning, it overcomes the limitations of existing UAV positioning technologies in complex environments. This invention uses a high-frequency dynamic infrared marker array as the positioning reference. This array consists of multiple independently controllable infrared light sources. By precisely controlling the flashing frequency of each infrared light source, it generates optical signals with unique temporal characteristics, overcoming the limitations of traditional static optical markers and providing UAVs with richer and more dynamic positioning information, thereby significantly improving positioning accuracy and reliability. Simultaneously, this invention innovatively combines an event camera with the high-frequency dynamic infrared marker array to form a highly efficient collaborative working mechanism. The event camera can capture changes in the optical event stream in real time. Utilizing its high-speed dynamic response characteristics, it accurately identifies the unique flashing frequency of each light source in the array, thereby estimating its own position. This combined mechanism of the invention provides all-weather, all-scene positioning capabilities. The strong penetrating power of the infrared light source enables stable operation in complex environments such as low light, nighttime, and foggy weather, while the high-speed dynamic response of the event camera ensures accurate capture of optical signals even in fast-moving scenarios.

[0013] Compared with existing technologies, this invention has several significant advantages. By utilizing a high-frequency scintillation infrared array and its unique encoding, combined with the high temporal resolution of an event camera, it enables high-precision detection and recognition of markers in complex environments, effectively resisting environmental interference and noise, and greatly improving the system's robustness. In high-speed movement and complex dynamic environments of UAVs, this invention maintains real-time, high-precision positioning capabilities, ensuring accurate navigation of UAVs in fast flight and complex scenarios, meeting the demands of high-dynamic operations. Furthermore, this invention solves the performance bottleneck problem of event cameras in low-light or static environments. By enhancing the signal through a high-frequency infrared array, it significantly enhances the system's robustness and adaptability, expanding its application scenarios.

[0014] Meanwhile, high-frequency infrared arrays are inexpensive and easy to deploy. Combined with event cameras, they can achieve a comprehensive positioning solution, avoiding the use of expensive, high-precision LiDAR and other complex equipment, thus reducing system cost and complexity. Furthermore, the event-driven data processing method significantly reduces computational resource consumption compared to traditional frame-driven vision solutions, making it suitable for resource-constrained embedded platforms and improving system real-time performance and energy efficiency. The system of this invention can respond to changes in the target environment within sub-millisecond timeframes, which is significantly superior to traditional frame-rate-based vision solutions, greatly improving the overall operational efficiency of the UAV.

[0015] Other beneficial effects of the embodiments of the present invention will be further described below. Attached Figure Description

[0016] Figure 1 is a block diagram of the nighttime precision landing system for unmanned aerial vehicles (UAVs) of the present invention.

[0017] Figure 2 is a system overall structure diagram of an embodiment of the present invention.

[0018] Figure 3 is a schematic diagram of system operation according to an embodiment of the present invention.

[0019] Figure 4 is a schematic diagram of the strobe infrared electronic control according to an embodiment of the present invention.

[0020] Figure 5 is a schematic diagram of PnP calculation according to an embodiment of the present invention. Detailed Implementation

[0021] The embodiments of the present invention will be described in detail below. It should be emphasized that the following description is merely exemplary and not intended to limit the scope and application of the present invention.

[0022] It should be noted that when a component is referred to as "fixed to" or "set on" another component, it can be directly on or indirectly on that other component. When a component is referred to as "connected to" another component, it can be directly connected to or indirectly connected to that other component. Furthermore, a connection can be used for fixing, coupling, or communication.

[0023] It should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0024] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of the present invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0025] This invention proposes an event-assisted nighttime precision landing system and method for unmanned aerial vehicles (UAVs). The invention uses a high-frequency dynamic infrared marker array as a positioning reference. By precisely controlling the independent flashing frequency of each infrared light source in the infrared marker array, an airborne event camera captures an optical event stream with unique temporal characteristics, thereby estimating the UAV's own position and achieving precise nighttime positioning and takeoff and landing control for the UAV.

[0026] Referring to Figures 1 to 3, this embodiment of the invention provides an event-assisted precise nighttime landing system for unmanned aerial vehicles (UAVs), including a high-frequency dynamic infrared marker array module, an event camera module, a pose calculation module, and a landing control module. The high-frequency dynamic infrared marker array module, which can be positioned near the UAV's landing target area, consists of multiple independently controllable infrared light sources and generates optical signals with unique temporal codes through a preset three-dimensional geometric layout. The event camera module, mounted on the UAV, captures the optical event stream of the infrared light sources and extracts the frequency features of the light sources based on high temporal resolution. The pose calculation module, connected to the event camera module, calculates the UAV's pose by combining the two-dimensional projection coordinates of the matching light source on the event camera's imaging plane with known three-dimensional spatial coordinates using the N-point perspective method (PnP). The landing control module controls the UAV's trajectory based on the real-time pose calculated by the pose calculation module, achieving precise nighttime landing. The system achieves sub-millisecond positioning feedback in complex environments through the coordinated operation of the independent light source flashing frequency of the infrared marker array module and the microsecond-level response characteristics of the event camera module.

[0027] Referring to Figure 4, in some embodiments, the high-frequency dynamic infrared marker array module includes an independent circuit driving unit: each infrared light source is connected to a transistor driving circuit; by precisely controlling the transistor's on / off timing, each light source can blink independently at a predetermined frequency; the output of the circuit driving unit is synchronized with the event capture rate of the event camera module. High-stability timing devices or periodic synchronization mechanisms can be used to compensate for clock drift to ensure signal decoding accuracy.

[0028] In some embodiments, the pose calculation module includes: a dynamic tracker unit: establishing a tracker for each matched light source with center point coordinates and search radius as attributes, and predicting the real-time position based on historical motion; a centroid update unit: clustering the event flow within the search radius and calculating the centroid, updating the center point coordinates through a low-pass filter; and a PnP calculation unit: performing pose calculation using at least four updated light source coordinates. Figure 5 illustrates the PnP calculation principle in detail.

[0029] In some embodiments, the PnP calculation unit is configured with a reprojection error verification mechanism: when the reprojection error of a certain light source exceeds the average distance threshold of the event centroid, the light source tracker is triggered to reinitialize; after the calculation is completed, a new round of PnP calculation is started immediately to realize asynchronous pose update.

[0030] In some embodiments, the UAV nighttime precision landing system further includes a multimodal fusion module, which fuses the event camera data, prior information on the UAV's motion state (such as inertial measurement unit (IMU) data), and optional visual sensor data, and optimizes the pose output through a Bayesian filter.

[0031] In some embodiments, the Bayesian filter of the multimodal fusion module adopts a Kalman filter or particle filter architecture, specifically including: a state space model construction unit: fusing PnP calculation results with IMU motion priors; a probability inference unit: suppressing event noise and pose jumps caused by high-speed motion; and an output interface: generating a smooth and optimized pose data stream.

[0032] Referring to Figures 1 to 3, this embodiment of the invention also provides a method for precise nighttime landing of a UAV based on the aforementioned UAV precise nighttime landing system, comprising the following steps: Step S1. Dynamic feature matching: The optical event stream of the infrared marker array is captured by the airborne event camera. Based on the time difference histogram analysis of the event stream, the unique frequency features of each light source are extracted to achieve light source identification and matching. In the initial stage, short-term observations can be used to cluster event groups and associate them with pre-stored light source features to achieve coarse matching, providing a basis for tracker creation. Step S2. Real-time pose calculation: The UAV pose is calculated by using the N-point perspective method (PnP) to solve the two-dimensional projection coordinates of the matched light source on the event camera imaging plane and the known three-dimensional spatial coordinates. Among them, the real-time position and attitude of the UAV are solved by using the known three-dimensional spatial coordinates of at least four matched light sources and their two-dimensional projections on the event camera imaging plane. Step S3. Landing control: The UAV trajectory is controlled according to the solved real-time pose of the UAV to achieve precise nighttime landing.

[0033] In some embodiments, step S1 specifically includes: filtering optical events from the event stream that conform to the characteristics of an infrared light source (this can be done by combining spatial region, wavelength characteristics, or event quantity thresholds to filter valid infrared events to reduce environmental interference); extracting the independent flicker frequency characteristics of each light source through a time difference histogram; and associating and matching the events with infrared light sources encoded with preset frequencies based on the frequency characteristics. The correspondence between the flicker frequency of the light source and the observed events can be established by separating event types or using digital encoding to avoid frequency confusion.

[0034] In some embodiments, step S2 specifically includes: establishing a dynamic tracker for the matching light source, predicting the center point and search radius of the imaging plane based on historical motion; clustering events within the search radius and calculating the centroid position each time the light source flashes, updating the center point coordinates through a low-pass filter; performing PnP calculation based on the updated center point coordinates, preferably also verifying the tracking effectiveness through reprojection error.

[0035] In some embodiments, the method further includes multimodal fusion pose optimization: a Bayesian filter is used to fuse the PnP calculation results with prior information about the UAV's motion state (such as IMU data) and optional visual sensor data. Noise is suppressed and pose estimation is smoothed through probabilistic inference. More preferably, the optimization of the Bayesian filter specifically includes: establishing a UAV motion state space model to fuse instantaneous observation data from the event camera with motion priors; using the PnP calculation results as observations and the motion state (such as IMU data) as priors; performing probabilistic inference through Kalman filtering or particle filtering to optimize pose estimation and suppress instantaneous jumps and event noise caused by high-speed motion. By combining event camera data, prior information about the UAV's motion state, and optional visual sensor data, multi-sensor fusion improves the robustness of positioning in complex environments.

[0036] This invention proposes an innovative scheme for precise take-off, landing, and positioning of unmanned aerial vehicles (UAVs), overcoming the shortcomings of existing nighttime positioning technologies for UAVs. It enables high-precision, high-reliability, and low-latency positioning in complex urban environments. The scheme employs a high-frequency dynamic infrared marker array as a positioning reference. By precisely controlling the independent flashing frequency of each infrared light source in the array, the event camera can capture optical event streams with unique temporal characteristics, thereby achieving precise identification and tracking of each light source and ultimately estimating the UAV's own position. This invention not only provides high-precision and high-reliability positioning capabilities in complex environments and at night, but also effectively solves problems such as signal obstruction, noise interference, high computational resource consumption, and difficulties in nighttime imaging in existing technologies, providing strong support for the safe and reliable operation of UAVs.

[0037] The following describes specific embodiments of the present invention.

[0038] An event-assisted nighttime precision landing system and method for unmanned aerial vehicles (UAVs) is disclosed. This technical solution for precise UAV takeoff, landing, and positioning utilizes a high-frequency dynamic infrared marker array as a positioning reference. By precisely controlling the independent flashing frequency of each infrared light source in the array, an event camera can capture optical event streams with unique temporal characteristics, thereby achieving precise identification and tracking of each light source. The infrared marker array employs a specific three-dimensional spatial geometric layout, with its coordinate parameters pre-calibrated and stored, providing a known reference framework for subsequent pose calculations. Based on the high temporal resolution of the event camera, the system can detect the flashing events of infrared light sources in real time and extract the characteristic frequencies of each light source through time difference histogram analysis, effectively overcoming the motion blur and illumination interference problems present in traditional visual positioning.

[0039] Referring to Figure 2, the workflow of the UAV nighttime precision landing system is as follows: A dynamic visual identifier (high-frequency infrared array) emits optical signals with unique time-series encoding; an event camera captures the signals and generates an optical event stream; the event stream is processed through key feature detection, extraction, and matching methods (including candidate identifier selection, flashing frequency recognition, and LED sequence matching); the matching results are output to a high-precision, low-latency positioning method (based on N-point perspective and Bayesian filter, performing state prediction, state update, and reprojection error verification); the positioning results are optimized by combining a multimodal fusion method (fusing data such as IMU motion state data and camera visual sensor data); finally, the UAV's 3D position information (lx, ly, lz) is output to achieve precise landing control.

[0040] The method for precise nighttime landing of UAVs mainly consists of two parts: key feature matching and high-precision, low-latency positioning based on perspective. An infrared stroboscopic array is used as the signal generator, and an airborne event camera is used as the signal receiver to achieve feature signal transmission and matching. In addition, a high-precision, low-latency UAV positioning method based on the N-point perspective method and Bayesian filters is designed, using reprojection errors to perform asynchronous spatial position estimation for the UAV.

[0041] The feature matching section employs a method for detecting, extracting, and matching key features based on dynamic identifiers and event cameras. This includes steps such as candidate identifier selection, flashing frequency identification, and high-frequency flashing infrared array matching. Specifically, this invention designs and arranges a high-frequency flashing infrared array, assigning it different flashing frequencies, thereby enabling the event camera to individually identify each infrared light. After identifying four or more fixed ground infrared signals, PnP calculation can be performed to estimate the UAV's own position and attitude.

[0042] The localization section uses a Bayesian filter to optimize and fuse the outputs of the event camera and the PnP algorithm. This filter establishes a state-space model of the UAV's motion, combining instantaneous observation data provided by the event stream with the system's motion priors. Through probabilistic inference, it effectively suppresses noise data and smoothly optimizes pose estimation. In practice, a Kalman filter or particle filter architecture is employed, flexibly configured according to computational resource requirements, improving positioning accuracy while ensuring real-time performance. This filtering mechanism is particularly suitable for handling random noise in event camera data and instantaneous positioning jumps caused by high-speed UAV movement, enabling the system to maintain stable positioning performance even in complex dynamic environments.

[0043] Figure 3 illustrates a schematic diagram of the system operation according to an embodiment of the present invention. The present invention uses a high-frequency scintillation infrared array as a signal generator, leveraging its unique time coding and frequency modulation characteristics to generate optical signals with distinct features. Simultaneously, an airborne event camera serves as a signal receiver, utilizing its high temporal resolution and event-driven characteristics to accurately capture the feature signals emitted by the infrared array. By matching the signals captured by the event camera with preset signal features, efficient transmission and accurate identification of the feature signals are achieved, thus providing a reliable technical foundation for the precise positioning and navigation of UAVs. This method not only improves the efficiency and accuracy of signal transmission but also enhances the system's anti-interference capability and adaptability in complex environments.

[0044] Figure 4 illustrates a circuit design for controlling the flashing of multiple infrared light sources, which can be used to achieve the precise take-off and landing and positioning technology for UAVs described in this invention. The circuit contains nine identical drive modules, each controlling the flashing of one infrared light source via a transistor. This circuit diagram implements the control of a high-frequency flashing infrared array. By precisely controlling the on / off state of each transistor, each infrared light source can be controlled to flash at a predetermined frequency. These flashing infrared light sources serve as dynamic visual identifiers, working in conjunction with an event camera to achieve precise positioning and navigation for the UAV. The different flashing frequencies of each infrared light source allow the event camera to individually identify and track each light source, thereby extracting key information for positioning.

[0045] Figure 5 illustrates the positioning principle based on the N-point perspective method (PnP). c X represents the origin of the camera coordinate system; c , Y c Z c Represents the X, Y, and Z axes of the camera coordinate system; p L p represents a point in three-dimensional space. i d represents a point on the image plane; i Point p i Distance to the camera; R and t represent the rotation matrix and translation vector, respectively; O w X represents the origin of the world coordinate system; w Y w Z wThe X, Y, and Z axes represent the world coordinate system. Using known three-dimensional spatial reference points—in this invention, the positions of the high-frequency flashing infrared array—and their two-dimensional projections onto the image plane, the UAV's pose is calculated. By minimizing the reprojection error between the predicted projection points and the actually observed two-dimensional points, the PnP algorithm can accurately estimate the UAV's position and attitude relative to the reference point array. This detection algorithm generates a corresponding tracker for each flashing infrared array. Each tracker is characterized by its frequency (fi), center point (ci = [xi, yi]), and radius (ri), and is predicted based on historical motion at each moment. During each infrared flash, the tracker's centroid is calculated using all events within the current radius, and the centroid position is updated using a low-pass filter. This dynamic update method can adjust the radius and center point position in real time, adapting to environmental changes and UAV motion. Next, this invention uses the center point information of the currently tracked high-frequency flashing infrared array to asynchronously estimate the UAV's spatial position information. To improve the algorithm's update rate, a new round of PnP calculations is initiated immediately after the previous PnP algorithm completes. In the PnP calculation process, the calculation of reprojection error is a crucial step. If the reprojection error of a tracker exceeds the average distance of the event center point, tracking failure is determined, a tracking loss signal is triggered, and the tracker will then be reinitialized.

[0046] The method and system proposed in this invention have broad application prospects. In the field of precision delivery by logistics drones, the active infrared tagging positioning method of this invention can provide drones with all-weather, highly robust positioning capabilities. In scenarios such as indoor warehouses and densely populated urban areas where satellite navigation signals are rejected or interfered with, this method ensures that drones achieve centimeter-level positioning accuracy through a pre-set infrared beacon network, providing reliable technical support for automated logistics delivery. Furthermore, this technology can be extended to multiple application areas such as industrial inspection, agricultural plant protection, and emergency rescue.

[0047] Compared to current positioning methods that rely on visible light markers or lidar, the technology of this invention has significant advantages in terms of environmental adaptability and cost-effectiveness. It can operate stably under extreme lighting conditions, is unaffected by stray light interference, and will not fail to recognize markers due to high-speed movement. In the military and security fields, the positioning method of this invention exhibits unique stealth combat capabilities. The invisible nature of infrared markers, combined with the passive sensing mechanism of event cameras, enables drones to complete precise take-off and landing and material delivery while remaining completely concealed.

[0048] In terms of civilian commercial applications, the modular design of this invention allows for rapid adaptation to different levels of drone platforms. By adjusting the deployment density and modulation parameters of the infrared markers, it can flexibly meet the diverse positioning needs of drones ranging from consumer-grade recreational drones to industrial-grade operational drones. The standardized hardware interface and programmable driver architecture adopted by the system significantly lower the barrier to technology promotion, creating favorable conditions for large-scale commercial applications.

[0049] In summary, this invention proposes a high-precision landing system and method for unmanned aerial vehicles (UAVs) based on the collaborative perception of dynamic infrared markers and event cameras. By utilizing the microsecond-level response characteristics of a high-frequency modulated infrared marker array and an event camera, it enables precise positioning and attitude estimation of UAVs in complex environments.

[0050] Compared with the prior art, the advantages of the present invention are as follows: 1. With the unique coding of the high-frequency scintillation infrared array and the high temporal resolution of the event camera, high-precision detection and recognition of the markings can be achieved in complex environments, significantly improving the system's robustness against environmental interference and noise.

[0051] 2. It can maintain real-time and high-precision positioning capabilities in high-speed movement and complex dynamic scenarios of drones, ensuring accurate navigation of drones during rapid flight and meeting the needs of highly dynamic operations.

[0052] 3. By enhancing the signal with a high-frequency infrared array, the performance bottleneck of the event camera in low light or static environments is effectively solved, significantly enhancing the system's robustness and environmental adaptability, and expanding application scenarios.

[0053] 4. High-frequency infrared arrays are inexpensive and easy to deploy. Combined with event cameras, they can form a complete positioning solution, avoiding the investment in expensive equipment such as high-precision lidar and reducing system cost and complexity.

[0054] 5. Compared with traditional frame-driven vision solutions, event-driven data processing significantly reduces computing resource consumption, making it suitable for resource-constrained embedded platforms and improving system real-time performance and energy efficiency.

[0055] 6. The system can respond to changes in the target environment in sub-millisecond time, with a response speed significantly better than traditional vision solutions based on frame rate, greatly improving the overall operational efficiency of the UAV.

[0056] The above description provides a further detailed explanation of the present invention in conjunction with specific / preferred embodiments, and it should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various substitutions or modifications can be made to these described embodiments without departing from the concept of the present invention, and all such substitutions or modifications should be considered within the scope of protection of the present invention. In the description of this specification, the reference to terms such as "an embodiment," "some embodiments," "preferred embodiment," "example," "specific example," or "some examples," etc., indicates that the specific features, structures, materials, or characteristics described in connection with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification and the features of different embodiments or examples. Although the embodiments of the present invention and their advantages have been described in detail, it should be understood that various changes, substitutions, and modifications can be made herein without departing from the scope of protection of the patent application.

Claims

1. An event-assisted nighttime precision landing system for unmanned aerial vehicles, characterized in that, include: The high-frequency dynamic infrared marker array module consists of multiple independently controllable infrared light sources, which generate optical signals with unique timing codes through a preset three-dimensional geometric layout. An event camera module, mounted on the UAV, is used to capture the optical event stream of the infrared light source and extract the frequency features of the light source based on its high temporal resolution. A pose calculation module, connected to the event camera module, is used to calculate the UAV pose by combining the two-dimensional projection coordinates of the matching light source on the event camera's imaging plane with the known three-dimensional spatial coordinates using the N-point perspective method (PnP). A landing control module controls the UAV's trajectory based on the solved real-time pose, enabling precise nighttime landing. Furthermore, the independent light source flashing frequency of the infrared marker array module and the microsecond-level response characteristics of the event camera module work together to achieve sub-millisecond-level positioning feedback in complex environments.

2. The system as described in claim 1, characterized in that, The high-frequency dynamic infrared marker array module includes an independent circuit driving unit: each infrared light source is connected to a transistor driving circuit; by precisely controlling the transistor's on / off timing, each light source can blink independently at a predetermined frequency; the output of the circuit driving unit is synchronized with the event capture rate of the event camera module.

3. The system as described in claim 1 or 2, characterized in that, The pose calculation module includes: a dynamic tracker unit: establishing a tracker for each matched light source with center point coordinates and search radius as attributes, and predicting the real-time position based on historical motion; a centroid update unit: clustering the event flow within the search radius and calculating the centroid, and updating the center point coordinates through a low-pass filter; and a PnP calculation unit: performing pose calculation using at least four updated light source coordinates.

4. The system as described in claim 3, characterized in that, The PnP calculation unit is configured with a reprojection error verification mechanism: when the reprojection error of a certain light source exceeds the average distance threshold of the event centroid, the light source tracker is triggered to reinitialize; after the calculation is completed, a new round of PnP calculation is started immediately to realize asynchronous pose update.

5. The system according to any one of claims 1 to 4, characterized in that, Also includes: The multimodal fusion module integrates the event camera data, prior information on the UAV's motion state, and optional visual sensor data, and optimizes the pose output through a Bayesian filter.

6. The system as described in claim 5, characterized in that, The Bayesian filter of the multimodal fusion module adopts a Kalman filter or particle filter architecture, specifically including: a state space model construction unit: fusing PnP calculation results with IMU motion priors; a probability inference unit: suppressing event noise and pose jumps caused by high-speed motion; and an output interface: generating a smooth and optimized pose data stream.

7. A method for precise nighttime landing of a drone based on the system of any one of claims 1 to 6, characterized in that, Includes the following steps: S1. Dynamic Feature Matching: The optical event stream of the infrared marker array is captured by an airborne event camera. Based on the time difference histogram analysis of the event stream, the unique frequency features of each light source are extracted to achieve the identification and matching of the light source. S2. Real-time pose calculation: The UAV pose is calculated by using the N-point perspective method (PnP) to solve the two-dimensional projection coordinates of the matched light source on the event camera imaging plane and the known three-dimensional spatial coordinates; S3. Landing control: The UAV trajectory is controlled according to the solved real-time pose to achieve precise night landing.

8. The method as described in claim 7, characterized in that, Step S1 specifically includes: filtering optical events that match the characteristics of infrared light sources from the event stream; extracting the independent flicker frequency characteristics of each light source through a time difference histogram; and associating and matching the events with infrared light sources encoded with preset frequencies based on the frequency characteristics.

9. The method as described in claim 7 or 8, characterized in that, Step S2 specifically includes: establishing a dynamic tracker for the matching light source, predicting the center point and search radius of the imaging plane based on historical motion; clustering events within the search radius and calculating the centroid position each time the light source flashes, updating the center point coordinates through a low-pass filter; performing PnP calculation based on the updated center point coordinates, and preferably also verifying the tracking effectiveness through reprojection error.

10. The method according to any one of claims 7 to 9, characterized in that, Also includes: Multimodal fusion pose optimization: A Bayesian filter is used to fuse the PnP calculation results with the prior information of the UAV's motion state and optional visual sensor data. Noise is suppressed and pose estimation is smoothed through probabilistic inference. Preferably, the optimization of the Bayesian filter specifically includes: establishing a UAV motion state space model to fuse instantaneous observation data from the event camera with motion priors; using the PnP calculation results as observations and the motion state as priors; and using Kalman filtering or particle filtering for probabilistic inference to optimize pose estimation and suppress instantaneous jumps and event noise caused by high-speed motion.

Images