Systems and methods for analyzing waveforms using pulse shape information
By generating count histograms and trial histograms, and using maximum likelihood estimation to analyze the pulse shape information of the LiDAR system, the distance walk problem caused by the avalanche histogram stacking effect is solved, achieving more accurate object detection and autonomous driving support.
Patent Information
- Application Number
- CN202280033553.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2021-03-08
- Filing Date
- 2022-01-28
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2042-01-28
AI Technical Summary
Photon-count-based LiDAR systems face challenges in reliably estimating the range and intensity of objects at various reflectivities, particularly due to inaccurate range walk and pulse intensity estimation caused by the stacking effect in the avalanche histogram.
By generating count histograms and trial histograms of timestamp values, the pulse shape information is analyzed using maximum likelihood estimation techniques to correct for distance walk effects, accurately estimate the intensity under pulse saturation, and provide robust distance and intensity estimates.
This improves the accuracy of distance and intensity estimation of LiDAR systems under high-intensity conditions, reduces errors in traditional methods, and enables more accurate object detection and autonomous driving decision support.
Smart Images

Figure CN117295976B_ABST
Abstract
Description
[0001] Cross-references and priority requirements
[0002] This patent document claims priority to U.S. Patent Application No. 17 / 194,391, filed March 8, 2021, the entire contents of which are incorporated herein by reference. Technical Field
[0003] This disclosure relates to systems and methods for analyzing waveforms using pulse shape information. Background Technology
[0004] Modern vehicles have at least one onboard computer and an internet / satellite connection. Software running on these onboard computers monitors and / or controls the operation of the vehicle. If the vehicle is autonomous, it needs to be able to predict the trajectories of agents / actors in the scene in order to make safe and effective behavioral decisions. These trajectories can be predicted using LiDAR data generated by a light detection and ranging (LiDAR) system coupled to the vehicle. The LiDAR system can include a photon-counting-based LiDAR system (e.g., a Geiger-mode-based sensor / detector array).
[0005] For photon-count-based LiDAR systems, reliable range and intensity estimation of objects under various reflectivities is a challenging problem. Each LiDAR system emits a light waveform and receives a waveform representing the light reflected from the surface of at least one object. As the pulse intensity of the received waveform increases, the stacking effect in the avalanche histogram of the LiDAR system's photodiode becomes more pronounced. The stacking effect causes range walk (i.e., high-intensity received light pulses are skewed at the beginning of the pulse, resulting in inaccurate sample measurements), which negatively impacts object range estimation. Summary of the Invention
[0006] This disclosure relates to systems and methods for operating a LiDAR system. The method includes: receiving a waveform representing light reflected from a surface of at least one object; generating multiple timestamp values for photon detection events triggered by pulses in the waveform; generating a count histogram of the timestamp values; inferring a test histogram from the count histogram (the test histogram representing the number of times the photodetector of the LiDAR system is available during the received waveform); generating an estimated ranging distance from the LiDAR system to at least one object and an estimated intensity value of a given pulse of the waveform based on the results of analyzing the count histogram and the test histogram; determining a location using the estimated ranging distance from the LiDAR system to at least one object; and generating a LiDAR dataset comprising data points defined by the location and the estimated intensity values. The LiDAR dataset can be used to control the operation of a vehicle (e.g., autonomous driving operation).
[0007] The estimated intensity value represents an estimate of the pulse's return intensity. In some cases, the estimated intensity value is set as the maximum likelihood estimate of the pulse intensity given the pulse shape, count histogram, and test histogram. The maximum likelihood estimate of the pulse intensity is selected from multiple possible intensity values, each determined using a linear combination of the count histogram and test histogram. Each intensity value is determined according to the following mathematical equation.
[0008]
[0009] Where s represents the span size of at least a portion of the counting histogram, f(i,τ=j) represents the known pulse shape of the waveform at bar index i, the waveform having a centroid at waveform offset τ, and n i Let k represent the test count at bar i, where i represents the bar index in the test histogram corresponding to the given pulse, and k i This represents the count of a given pulse detection event at bar index i.
[0010] The estimated distance is set as the maximum likelihood estimate of the distance given multiple possible intensity values. In some cases, the maximum likelihood estimate of the distance is determined by performing a comparison operation, in which a count histogram is compared with multiple calculated reference histograms given specific conditions. Each comparison operation produces a probability score, which represents the timestamp value determined by the parameter τ. j and λ j The probability of a given possible value. The probability fraction is defined by the following mathematical equation.
[0011]
[0012] Where λ represents the previously calculated intensity value λ. jAnd τ represents a timestamp value, which is set to the given previously calculated intensity value λ. j The associated predefined value j.
[0013] The implementation system includes: a processor; and a non-transitory computer-readable storage medium, including programming instructions configured to enable the processor to implement the methods described above. Attached Figure Description
[0014] This solution will be described with reference to the following figures, in which the same numbers represent the same items in all figures.
[0015] Figure 1 It is a diagram illustrating the system.
[0016] Figure 2 It is an illustration of the structure used for vehicles.
[0017] Figure 3 This is an illustration of the LiDAR system.
[0018] Figure 4 It is an illustration of a computing device.
[0019] Figure 5 A block diagram is provided to help understand how vehicle control is implemented according to this solution.
[0020] Figures 6A-6B (Hereinafter referred to as “Figure 6”) provides a flowchart of an illustrative method for controlling autonomous vehicles.
[0021] Figure 7 A flowchart illustrating a method for generating LiDAR datasets according to this solution is provided.
[0022] Figure 8 A graph is provided demonstrating the ability of this solution to estimate the distance to an object (which is expected to be at the bar (bin) 50 corresponding to the pulse center) with increased intensity without bias.
[0023] Figure 9 A graph is provided showing that this solution can analyze waveforms even when up to half of the pulses are clipped.
[0024] Figure 10 Illustrative count histograms and illustrative test histograms are provided.
[0025] Figure 11 An illustrative reference histogram is provided and is useful for understanding the likelihood score of this solution and the role of the reference histogram suggestions.
[0026] Figure 12A , Figure 12B and Figure 12C (Collectively referred to as “Figure 12”) provides a diagram illustrating the effects of distance travel. Detailed Implementation
[0027] As used in this document, the singular forms “a,” “an,” and “the” include plural references unless the context clearly indicates otherwise. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. As used in this document, the term “comprising” means “including, but not limited to,” other terms relevant to this document. Definitions of other terms related to this document are included at the end of this specific embodiment.
[0028] "Electronic device" or "computing device" refers to a device that includes a processor and memory. Each device may have its own processor and / or memory, or the processor and / or memory may be shared with other devices, as in a virtual machine or container arrangement. The memory will contain or receive programming instructions that, when executed by the processor, cause the electronic device to perform one or more operations according to the programming instructions.
[0029] The terms “memory,” “memory device,” “data storage,” “data storage facility,” etc., refer to non-transitory means of means thereon for storing computer-readable data, programming instructions, or both. Unless otherwise specified, the terms “memory,” “memory device,” “data storage,” “data storage facility,” etc., are intended to include a single device embodiment, an embodiment in which multiple memory devices together or jointly store a set of data or instructions, and individual sectors within such devices.
[0030] The terms “processor” and “processing device” refer to the hardware component of an electronic device configured to execute programmed instructions. Unless otherwise specified, the singular terms “processor” or “processing device” are intended to include embodiments of a single processing device and embodiments in which multiple processing devices together or jointly perform processing.
[0031] The term "vehicle" refers to any form of transport capable of carrying one or more human occupants and / or goods and powered by any form of energy. The term "vehicle" includes, but is not limited to, automobiles, trucks, vans, trains, autonomous vehicles, airplanes, and drones. An "autonomous vehicle" is a vehicle with a processor, programmed instructions, and a transmission system that can be controlled by the processor without the need for a human operator. An autonomous vehicle can be fully autonomous, as it requires no human operator for most or all driving conditions and functions, or it can be semi-autonomous, as a human operator may be required under certain conditions or for certain operations, or a human operator may have control over the vehicle's autonomous driving system and be able to control the vehicle.
[0032] In this document, when terms such as “first” and “second” are used to modify nouns, this use is intended only to distinguish one item from another and is not intended to require a sequential order, unless otherwise specified. Additionally, when used, relative positional terms such as “vertical” and “horizontal” or “front” and “back” are intended relative to each other and do not need to be absolute, and only indicate one possible position of the device associated with these terms, depending on the orientation of the device.
[0033] As described above, reliable range and intensity estimation of objects at various reflectivities is a challenging problem for photon-counting-based LiDAR systems (e.g., Geiger-mode-based sensor / detector arrays). Each LiDAR system emits a light waveform and receives a waveform representing the light reflected from the surface of at least one object. As the pulse intensity of the received waveform increases, the stacking effect in the avalanche histogram of the photodiodes of the LiDAR system becomes more pronounced. The stacking effect causes range walk (i.e., high-intensity received light pulses are skewed at the beginning of the pulse, resulting in inaccurate sample measurements), which negatively impacts object range estimation. Furthermore, estimating the intensity of the received pulses can be particularly challenging due to shot noise in photon-triggered avalanche counting and the possibility of photodetector saturation under certain operating conditions. These problems are addressed by this solution.
[0034] Figure 12 illustrates the effect of distance walk. In Figure 12, the count histograms observed from the received rectangular pulses (shown as dashed lines) are shown when the received intensities are 0.1, 1, and 10 photons per detector per pulse. At an intensity of 0.1, the probability of avalanches (counts) is more uniform along the pulse span; therefore, the received count histogram is closer to a rectangular shape (in fact, the earlier bars have a slightly higher chance of avalanches than the later bars, but the difference is small, so the decay of the avalanche probability is not significant, and it appears as if the avalanche probability is uniform across the pulse span). At an intensity of 1, the avalanche probability of the earlier bars appears to be higher than that of the later bars, and this is due to the detector's deactivation after the avalanche event. Therefore, the histogram is more skewed to the left (the earlier bars). This skewness is even stronger at an intensity of 10, where most of the avalanche probability is in the first bar of the received pulse under this high-intensity condition, and much smaller in the later bars.
[0035] This solution provides a system and method for analyzing waveforms using pulse shape information. Waveform analysis corrects for the effects of range walk, accurately estimates intensity under pulse saturation, and provides robust distance and intensity estimates with clipped pulses. Pulse saturation and clipped pulses occur when too much light is received by a LiDAR system. Distance and intensity estimation are based on maximum likelihood estimation techniques. Compared to the ranging method in this solution, conventional ranging methods such as centroid, zero-crossing, and constant fractional discriminators have lower accuracy because they do not fully utilize pulse shape information.
[0036] Figure 8 The figure illustrates the ability of this solution to estimate the distance to an object (which is expected to be at bin 50 corresponding to the pulse center) with increased intensity without bias. Figure 8 The table shows histograms for received intensities of 6, 11, 21, 30, and 41 photons per pulse. The vertical lines indicate the pulse center that the proposed algorithm aims to estimate. Crosses represent the maximum points in the received count histograms, and the x-coordinate of points closer to the vertical lines represents the estimated pulse center. It can be seen that the estimation is consistent across different received intensities and the skewness of the count histograms.
[0037] Figure 9 The figure illustrates that this solution is able to analyze waveforms even when up to half of the pulses are clipped. More specifically, Figure 9 The results of the proposed algorithm are shown when the analytical portion of the received count histogram is contained within a rectangular box. It demonstrates that the algorithm is able to successfully estimate the appropriate offset and intensity of the pulse if the box contains more than half of the waveform. The expected received intensity is eight photons per pulse. Figure 9Four overlap conditions are shown. The first condition has just less than half of the pulses, so the estimated intensity is almost half of the actual intensity. As shown in the second, third, and fourth conditions, once the box contains more than half of the waveform, the corresponding intensity and distance estimates become consistent, regardless of the percentage of the captured waveform (as long as it is more than half).
[0038] This method typically includes operations for analyzing the LiDAR returned waveform to more accurately estimate the pulse intensity and the ranging distance to the object. The operations include: emitting light (e.g., 100 light pulses) from the LiDAR system; receiving a waveform representing the light reflected from the surface of at least one object by a photon-counting sensor of the LiDAR system (e.g., a Geiger-mode-based sensor / detector array); generating a timestamp value for each detection of pulses in the received waveform (each timestamp value representing the time of flight of a given pulse in the emitted light waveform); generating a count histogram of the timestamp values (e.g., plotting the timestamp values); inferring a trial histogram from the count histogram, which represents how many times a given photodetector of the photon-counting sensor is available during the pulse detection of the received waveform; and analyzing the count histogram and the trial histogram to generate an estimated ranging distance d to the object and an estimated intensity value λ.
[0039] The estimated intensity value λ represents an estimate of the returned intensity (amplitude) of the laser pulse and is set to the maximum likelihood estimate of the pulse intensity given a known pulse shape, count histogram, and test histogram. The returned intensity of the pulse depends on the reflectivity of the object. Therefore, this solution involves calculating multiple possible intensity values λ for a given known pulse shape. j , where j is a predefined integer (e.g., 1-20). Each intensity value λ is generated by implementing a linear combination of the count histogram and the trial histogram. j .
[0040] The estimated ranging distance d represents the estimated distance between the LiDAR system and the object. The estimated ranging distance d is set to be equal to a given number of possible intensity values λ. j The maximum likelihood estimate of the ranging distance is typically determined by comparing a count histogram with each of several computed reference histograms given certain conditions, and identifying the conditions under which the count histograms result in the best match measured by the probability score Lj. Each comparison operation can produce a probability score Lj, which represents the timestamp value of the measurement determined by the parameter τ. j and λ j The probability caused by a given possible value.
[0041] This operation also involves: selecting τ with the maximum probability score Lj associated with it. j and λ jThe value of τ is set; the estimated distance d is set to be equal to the selected τ. j The value is set, and the estimated intensity value λ is set to be equal to the selected λ. j The system calculates the position (x, y, z coordinates) using the LiDAR system's location and orientation, the LiDAR system's angle, and the estimated ranging distance τ to the object; repeats the above operation for each LiDAR data point; and generates a LiDAR dataset defining a point cloud. Each data point in the point cloud has the following associated attributes: position (x, y, z coordinates); and intensity value. In some scenarios, the LiDAR dataset may be optionally used to control the operation of autonomous vehicles (e.g., for feature detection / extraction and / or LiDAR point classification (e.g., ground, vegetation, buildings, water, railways, road surfaces, power lines, bridges, vehicles, etc.)).
[0042] This document describes the solution within the context of autonomous vehicles. This solution is not limited to autonomous vehicle applications. It can be used in other applications such as robotics, artificial intelligence (AI), measurement, and / or system performance.
[0043] Explanatory System
[0044] Now for reference Figure 1 The illustration provides a diagram of an illustrative system 100. System 100 includes a vehicle 1021 that travels along a road in a semi-autonomous or autonomous driving mode. The vehicle 1021 may be an AV. The AV 1021 may include, but is not limited to, land vehicles (such as...) Figure 1 (as shown), aircraft or ships.
[0045] AV 1021 is typically configured to detect objects 1022, 114, and 116 in its vicinity. Objects may include, but are not limited to, vehicles 1022, cyclists 114 (such as riders of bicycles, e-scooters, motorcycles, etc.), and / or pedestrians 116. For example, object detection can be performed by analyzing sensor data generated by at least one sensor device on AV 1021 and / or information received via communication links 150, 152, and 154 from the object's communication device (e.g., transceiver, beacon, and / or smartphone). Communication links 150, 152, and 154 may include, but are not limited to, V2X communication links. The term "V2X" refers to communication between a vehicle and any entity that may affect or be affected by the vehicle. This information may include, but is not limited to, information specifying actions or operations that the object has performed, is performing, and / or will perform.
[0046] When performing such detection, AV 1021 performs the following operations: generating one or more possible object trajectories for the detected object; determining one or more object classifications, targets, and / or future intentions of the detected object; and using at least one of the generated possible object trajectories, object classifications, targets, and future intentions to facilitate the determination of a vehicle trajectory. AV 1021 can then perform operations to follow the vehicle trajectory.
[0047] In some cases, AV 1021 performs additional actions to determine whether there is an unwanted risk level of collision between the AV and the object within a threshold time period (e.g., 1 minute). If so, if AV 1021 follows the vehicle's trajectory and performs any of a plurality of dynamically generated emergency maneuvers within a predefined time period (e.g., N milliseconds), AV 1021 performs an action to determine whether a collision can be avoided. If a collision can be avoided, AV 1021 takes no action, or optionally performs cautious maneuvers (e.g., gently decelerating). Conversely, if a collision cannot be avoided, AV 1021 immediately performs emergency maneuvers (e.g., braking and / or changing direction).
[0048] Now for reference Figure 2 It provides an illustration of a system architecture 200 for vehicles. Figure 1 Vehicles 1021 and / or 1022 may have the same Figure 2 The system architecture shown is the same as or similar to the system architecture shown. Therefore, the following discussion of system architecture 200 is sufficient for understanding. Figure 1 The vehicles are 1021 and 1022.
[0049] like Figure 2 As shown, vehicle 200 includes an engine or motor 202 and various sensors 204-218 for measuring various parameters of the vehicle. In a gasoline-powered or hybrid vehicle with a fuel-powered engine, for example, sensors may include an engine temperature sensor 204, a battery voltage sensor 206, an engine revolutions per minute (RPM) sensor 208, and a throttle position sensor 210. If the vehicle is an electric or hybrid vehicle, it may have an electric motor and therefore will have sensors such as a battery monitoring system 212 (to measure battery current, voltage, and / or temperature), a motor current sensor 214, and a voltage sensor 216, as well as motor position sensors such as a resolver and encoder 218.
[0050] For example, common operating parameter sensors for both types of vehicles include: position sensors 236, such as accelerometers, gyroscopes, and / or inertial measurement units; speed sensors 238; and odometer sensors 240. The vehicle may also have a clock 242, which the system uses to determine the vehicle's time during operation. Clock 242 can be encoded into an onboard computing device; it can be a separate device, or multiple clocks may be available.
[0051] The vehicle will also include various sensors that operate to collect information about the environment in which the vehicle is traveling. For example, these sensors may include: a position sensor 260 (e.g., a Global Positioning System (GPS) device); object detection sensors, such as one or more cameras 262; a LiDAR sensor system 264; and / or a radar and / or sonar system 266. Sensors may also include environmental sensors 268, such as precipitation sensors and / or ambient temperature sensors. Object detection sensors enable the vehicle to detect objects within a given distance range of the vehicle 200 in any direction, while environmental sensors collect data about environmental conditions within the vehicle's operating area.
[0052] During operation, information is transmitted from sensors to onboard computing unit 220. Onboard computing unit 220 analyzes the data captured by the sensors and optionally controls the operation of the vehicle based on the results of the analysis. For example, onboard computing unit 220 may: control braking via brake controller 232; control direction via steering controller 224; control speed and acceleration via throttle controller 226 (in gasoline-powered vehicles) or motor speed controller 228 (such as a current level controller in electric vehicles); control differential gear controller 230 (in vehicles with a transmission); and / or control other controllers.
[0053] Geographic location information can be transmitted from location sensor 260 to onboard computing device 220, which can then access a map of the environment corresponding to the location information to determine known fixed features of the environment, such as streets, buildings, parking signs, and / or stop / pass signals. Images captured from camera 262 and / or object detection information captured from sensors such as LiDAR system 264 are transmitted from those sensors to onboard computing device 220. The object detection information and / or captured images are processed by onboard computing device 220 to detect objects near vehicle 200. Any known or to be known techniques for object detection based on sensor data and / or captured images may be used in the embodiments disclosed in this document.
[0054] LiDAR information is transmitted from LiDAR system 264 to onboard computing device 220. Additionally, captured images are transmitted from camera 262 to onboard computing device 220. The LiDAR information and / or captured images are processed by onboard computing device 220 to detect objects near vehicle 200.
[0055] When the onboard computing device 220 detects a moving object, it generates one or more possible object trajectories and analyzes them to assess the risk of a collision between the object and the AV. If the risk exceeds an acceptable threshold, the onboard computing device 220 performs an operation to determine whether a collision can be avoided if the AV follows a defined vehicle trajectory and / or performs one or more dynamically generated emergency maneuvers within a predefined time period (e.g., N milliseconds). If a collision can be avoided, the onboard computing device 220 can cause the vehicle 200 to perform cautious maneuvers (e.g., gently decelerate, accelerate, or turn). Conversely, if a collision cannot be avoided, the onboard computing device 220 will cause the vehicle 200 to take emergency maneuvers (e.g., brake and / or change direction).
[0056] Now for reference Figure 3 An illustrative diagram of the LiDAR system 300 is provided. Figure 2 The LiDAR system 264 can be the same as or substantially similar to the LiDAR system 300. Therefore, a discussion of the LiDAR system 300 is sufficient for understanding. Figure 2 The LiDAR system 264.
[0057] like Figure 3 As shown, the LiDAR system 300 includes a housing 306 that can rotate 360° about a central axis, such as a hub or axle 316. The housing may include a transmitter / receiver aperture 312 made of a light-transparent material. Although in Figure 3 A single aperture is shown, but this solution is not limited in this respect. In other cases, multiple apertures may be provided for emitting and / or receiving light. In either case, as the housing 306 rotates about its internal components, the LiDAR system 300 can emit light through one or more apertures 312 and receive reflected light returning toward one or more apertures 211. In an alternative scenario, the outer shell of the housing 306 may be a stationary dome, at least partially made of a light-transparent material, with rotatable components inside the housing 306.
[0058] Within a rotating housing or stationary dome is a light emitter system 304, configured and positioned to generate and emit waveforms (including light pulses, e.g., 100 light pulses) via one or more laser emitter chips or other light-emitting devices, through aperture 312 or through the transparent dome of housing 306. Emitter system 304 may include any number of individual emitters (e.g., 8 emitters, 64 emitters, or 128 emitters). Emitters may emit light of substantially the same or different intensities. Individual beams emitted by light emitter system 304 will have distinct polarization states that are not uniform across the array. As an example, some beams may have vertical polarization, while others may have horizontal polarization. The LiDAR system will also include one or more photodetectors 308, each containing a photodetector or photodetector array, positioned and configured to receive light reflected back into the system. The photodetector array may include, but is not limited to, a Geiger-mode avalanche photodiode (GmAPD) array. The GMAPD array can detect a single photon and generate an electrical pulse of sufficient amplitude to directly trigger a logic event in its coupled readout circuitry. The readout circuitry digitally counts or timestamps the photon detection events (i.e., when a logic event is triggered by an electrical pulse generated by the GMAPD array). The readout circuitry may include, but is not limited to, complementary metal-oxide-semiconductor (CMOS) circuitry. The light emitter system 304 and photodetector 308 will rotate with the rotating housing, or they will rotate within the stationary dome of the housing 306. One or more optical element structures 310 may be positioned in front of the light emitter system 304 and / or photodetector 308 to serve as one or more lenses or waveplates that focus and guide light passing through the optical element structures 310.
[0059] One or more optical element structures 310 may be positioned in front of mirror 312 to focus and guide light passing through the optical element structure 310. As shown below, the system includes optical element structures 310 positioned in front of mirror 312 and connected to a rotating element of the system, such that the optical element structures 310 rotate together with mirror 312. Alternatively or additionally, the optical element structure 310 may include multiple such structures (e.g., lenses and / or waveplates). Optionally, multiple optical element structures 310 may be arranged in an array on or integrally formed with the housing portion of housing 306.
[0060] Optionally, each optical element structure 310 may include a beam splitter that separates the light received by the system from the light generated by the system. For example, the beam splitter may include a quarter-wave or half-wave plate to perform the separation and ensure that the received light is directed to the receiver unit rather than the transmitter system (this could happen without such a wave plate, since the emitted and received light should exhibit the same or similar polarization).
[0061] The LiDAR system will include a power supply unit 318, a motor 316, and electronic components for powering the light-emitting unit 304. The LiDAR system will also include an analyzer 314 having elements such as a processor 322 and a non-transitory computer-readable medium 320 containing programming instructions configured to enable the system to receive data collected by the photodetector unit, analyze the data to measure the characteristics of the received light, and generate information that a connected system can use to make operational decisions regarding the environment from which the data is collected. Optionally, the analyzer 314 may be integrated with the LiDAR system 300, as shown, or some or all of the analyzers may be external to the LiDAR system and communicatively connected to the LiDAR system via a wired or wireless communication network or link.
[0062] Now for reference Figure 4 It provides an illustration of the architecture for the computing device 400. Figure 1 The computing device 110 and / or Figure 2 The onboard computing device 220 is the same as or similar to the computing device 400. Therefore, a discussion of the computing device 400 is sufficient to understand... Figure 1 The computing device 110 and Figure 2 The airborne computing device 220.
[0063] The computing device 400 may include a ratio Figure 4 The components shown may include more or fewer components. However, the components shown are sufficient to disclose an illustrative solution for implementing this solution. Figure 4 The hardware architecture represents an implementation of a representative computing device configured to operate a vehicle, as described herein. Thus, Figure 4 The computing device 400 implements at least a portion of the method described herein.
[0064] Some or all of the components of the computing device 400 may be implemented as hardware, software, and / or a combination of hardware and software. Hardware includes, but is not limited to, one or more electronic circuits. Electronic circuits may include, but are not limited to, passive components (e.g., resistors and capacitors) and / or active components (e.g., amplifiers and / or microprocessors). Passive and / or active components may be adapted, arranged, and / or programmed to perform one or more of the methods, processes, or functions described herein.
[0065] like Figure 4As shown, computing device 400 includes a user interface 402, a central processing unit (CPU) 406, a system bus 410, a memory 412 connected to and accessible by other parts of computing device 400 via the system bus 410, a system interface 460, and a hardware entity 414 connected to the system bus 410. The user interface may include input and output devices that facilitate user-software interaction for controlling the operation of computing device 400. Input devices include, but are not limited to, a physical and / or touch keyboard 450. Input devices may be via wired or wireless connections (e.g., The system interface 460 is connected to the computing device 400. Output devices include, but are not limited to, a speaker 452, a display 454, and / or a light-emitting diode 456. The system interface 460 is configured to facilitate wired or wireless communication to and from external devices, such as network nodes such as access points.
[0066] At least some of the execution of hardware entity 414 involves actions involving accessing and using memory 412, which may be random access memory (RAM), a disk drive, flash memory, optical disc read-only memory (CD-ROM), and / or another hardware device capable of storing instructions and data. Hardware entity 414 may include a disk drive unit 416 that includes a computer-readable storage medium 418 on which one or more sets of instructions 420 (e.g., software code) are stored, configured to implement one or more of the methods, processes, or functions described herein. Instructions 420 may also reside wholly or at least partially within memory 412 and / or CPU 406 during their execution by computing device 400. Memory 412 and CPU 406 may also constitute machine-readable media. As used herein, the term "machine-readable media" refers to a single medium or multiple media (e.g., a centralized or distributed database and / or associated caches and servers) storing one or more sets of instructions 420. As used herein, the term "machine-readable medium" also means any medium capable of storing, encoding, or carrying a set of instructions 420 for execution by computing device 400 and causing computing device 400 to perform any one or more methods of this disclosure.
[0067] Now for reference Figure 5 A block diagram is provided for understanding how vehicle control is implemented according to this solution. All operations performed in blocks 502-510 can be controlled by the vehicle's onboard computing device (e.g., Figure 1 Execute AV 1021.
[0068] In box 502, the position of the vehicle is detected. This can be based on data from the vehicle's position sensors (e.g., ...). Figure 2The detection is performed using sensor data output from the location sensor 260. This sensor data may include, but is not limited to, GPS data. The detected vehicle location is then transmitted to box 506.
[0069] In box 504, objects are detected near the vehicle. This is based on data from the vehicle's camera (e.g., ...). Figure 2 Camera 262), LiDAR system (e.g., Figure 2 The detection is performed using sensor data output from the LiDAR system 264 and / or radar / sonar system 264. Techniques for detecting objects are well known in the art. Any known or to be known object detection techniques may be used herein. Information 512 regarding the detected object is passed to box 506. This information 512 includes, but is not limited to, the trajectory (or spatial description) of the detected object. The trajectory (or spatial description) may include at least one predicted trajectory or path of the object, the object's velocity, the object's full range, the object's direction of travel, the object's classification, the object's target, and / or the object's predicted future intent. The predicted object trajectory or path may include, but is not limited to, a straight path or a curved path pointing towards the lane direction.
[0070] The object detection information 512 output from box 504 can then be used to facilitate at least one autonomous driving operation (e.g., feature detection / extraction, LiDAR point classification (e.g., ground, vegetation, buildings, water, railways, roads, surfaces, power lines, bridges, vehicles, etc.), object tracking, object trajectory prediction, vehicle trajectory determination, and / or collision avoidance). For example, the object's current position, the predicted object trajectory, and the object's predicted future intent can be used to determine the vehicle trajectory in box 506, and / or trigger cautious or emergency maneuvers in box 506. This solution is not limited to the details of this example.
[0071] In box 506, a vehicle trajectory is generated using information from boxes 502 and 504. Techniques for determining vehicle trajectories are well-known in the art. For example, in some scenarios, such techniques involve determining the trajectory of an AV when an object is in front of the AV, the object has a heading aligned with the direction the AV is moving, the object is classified as a functionally stopped vehicle, the object is associated with a lane-blocking target, and the object has a predicted future stopping intention; the trajectory will pass over the object in these situations. This solution is not limited to the details of this scenario. A vehicle trajectory 50 can be determined based on location information from box 52, object detection information from box 54, and map information 58 (which is pre-stored in the vehicle's data storage). The vehicle trajectory 50 can represent a smooth path without sudden changes that would cause discomfort to passengers. For example, the vehicle trajectory is defined by a travel path along a given lane of a road in which the object is not predicted to travel for a given amount of time. The vehicle trajectory 520 is then provided to box 508.
[0072] In box 508, steering angle and speed commands are generated based on the vehicle trajectory 520. The steering angle and speed commands are provided to box 510 for vehicle dynamic control.
[0073] Referring now to Figure 6, a method for controlling vehicles (e.g., Figure 1 A flowchart illustrating method 600 of a vehicle 1021. At least a portion of method 600 is comprised of an onboard computing device (e.g., Figure 2 The onboard computing device 220 executes this. For each object (e.g., one that has been detected within a certain distance range from the vehicle at any given time)... Figure 1 Transportation 1022 Figure 1 114 cyclists and / or Figure 1 Pedestrian 116) Execute method 600.
[0074] Method 600 includes multiple operations 602-630. This solution is not limited to the specific order of operations 602-630 shown in Figure 6. For example, operation 620 can be executed in parallel with operations 604-618, rather than after them as shown in Figure 6.
[0075] like Figure 6A As shown, method 600 begins at 602 and continues to 604, where the vehicle trajectory of the AV is generated (e.g., Figure 5 The vehicle trajectory (520) can represent a smooth path without sudden changes that would cause discomfort to passengers, or it can represent emergency maneuvers to avoid a collision. Techniques for determining vehicle trajectories are well known in the art. In some scenarios, it is based on position sensors (e.g., AV) Figure 2 The position information generated by the position sensor 260, and the position information generated by the AV's onboard computing device (e.g., Figure 2 Object detection information generated by the onboard computing device 220 of the AV, and by at least one camera of the AV (e.g., Figure 2 Images captured by camera 262) and generated by at least one LiDAR system (e.g., Figure 2 The LiDAR dataset generated by the LiDAR system 264, and the memory stored in the AV (e.g., Figure 4 The map information and / or lane information in the memory 412) are used to determine the vehicle trajectory.
[0076] Once the vehicle trajectory is generated, method 600 continues to 605, where the AV performs operations to detect objects in its vicinity. Object detection algorithms are well known in the art. Object detection is then used to facilitate at least one autonomous driving operation (e.g., object tracking operation, object trajectory prediction operation, vehicle trajectory determination operation, and / or collision avoidance operation).
[0077] Therefore, method 600 continues to 606, where for the object detected in 605 (e.g., Figure 1 (Vehicle 1022, cyclist 114, or pedestrian 116) determine one or more possible object trajectories (e.g., Figure 5 Possible object trajectories 512). Techniques for determining the trajectory of a detected object are well known. In some scenarios, sensor data is used to determine the object trajectory. Alternatively or additionally, other information may be used to determine the object trajectory. This other information may include, but is not limited to, information transmitted via a wireless communication link (e.g., Figure 1 The links 150, 152, and 154 receive information from the object's communication device (e.g., transceiver, beacon, and / or smartphone). This information may specify actions or operations that the object has performed, is performing, and / or will perform (e.g., right / left turn, lane change, acceleration / deceleration, driving path, etc.).
[0078] Next, in 610, it is determined whether the vehicle trajectory generated in 604 and the possible object trajectory generated in 606 intersect each other. If they do not intersect [611: No], then 612 is executed, where method 600 returns to 604.
[0079] Conversely, if they do intersect [611: Yes], method 600 proceeds to 614, where a time value is determined. This time value represents the time at which a collision will occur if the AV follows the vehicle trajectory and the object (e.g., it has been classified as a driving vehicle and has driving intent) follows a possible object trajectory. The time value determined in 614 is then compared to a threshold time value, as shown in 616. A threshold time value (e.g., one second or more seconds) is selected based on the given application. If the time value is greater than the threshold time value [616: No], 618 is executed, where method 600 returns to 604. If the time value is equal to or less than the threshold time value [616: Yes], method 600 proceeds to 620-622. 620-622 involves: dynamically generating one or more emergency maneuver profiles based on the vehicle trajectory and possible object trajectories / predicted driving paths; and determining whether a collision can be avoided if the AV follows the vehicle trajectory and performs any of the emergency maneuvers within a predefined time period (e.g., N milliseconds). Upon completion of 622, method 600 proceeds to Figure 6B 624.
[0080] Now for reference Figure 6B If a collision cannot be avoided within a predefined time period [624: No], then 626 is executed, whereby the AV immediately takes emergency maneuver. Emergency maneuver may include, but is not limited to, one of the dynamically generated emergency maneuvers discussed above with respect to 620. Techniques for taking emergency maneuver of the AV are well known in the art. Any known or to be known techniques for taking emergency maneuver of the AV may be used herein. Subsequently, 5630 is executed, where method 600 terminates or other processing is performed.
[0081] Conversely, if a collision can be avoided within a predefined time period [624: Yes], then 628 is executed, whereby the AV is optionally subjected to careful maneuvering (e.g., gentle deceleration). Techniques for causing the AV to perform careful maneuvering (such as deceleration) are well known in the art. Any known or to be known techniques for causing the AV to perform careful maneuvering may be used herein. Subsequently, 630 is executed, where method 600 terminates or other processing is performed.
[0082] Now for reference Figure 7 A flowchart of an illustrative method 700 for generating LiDAR datasets is provided. Method 700 may be at least partially generated by a processor (e.g., Figure 3 Processor 322 and / or Figure 4 CPU 406), LiDAR system (e.g., Figure 2 LiDAR system 264 and / or Figure 3 LiDAR system 300) and / or computing device (e.g., Figure 1 Computing device 110 Figure 2 Onboard computing device 220 and / or Figure 4 The computing device 400 performs the operation. LiDAR datasets can be used in various applications. For example, LiDAR datasets can be used to control vehicles (e.g., Figure 1 The operations of the AV (1021) (e.g., via object detection, object classification, feature detection / extraction, and / or LiDAR point classification (e.g., ground, vegetation, buildings, water, railways, roads, surfaces, power lines, bridges, vehicles, etc.)). Thus, the LiDAR dataset can be used at least in box 605 of Figure 6 to detect objects approaching the AV and / or in box 606 to generate trajectories of the detected objects.
[0083] like Figure 7 As shown, method 700 begins at 702 and continues to 704, wherein the LiDAR system (e.g., Figure 2 The LiDAR system 264) light emitter system (e.g., Figure 3 The light emitter system 304 emits light. The light is emitted in the form of a waveform including a pulse sequence (e.g., 100 light pulses at 1 ms intervals). In 706, the photodetector of the LiDAR system (e.g., Figure 3 The photodetector 308 receives a waveform representing the reflected light. The photodetector may include one or more photodetectors positioned and configured to receive light from at least one object (e.g., Figure 1 Transportation 1022 Figure 1 114 cyclists and / or Figure 1 The light reflected back to the LiDAR system by a pedestrian (116). The photodetector may include, but is not limited to, a GmAPD detector, which detects photons and generates electrical pulses of sufficient amplitude to directly trigger logic events in its coupled readout circuitry. The readout circuitry digitally counts and / or timestamps the photon detection events. Therefore, N timestamp values are generated in 708. Each timestamp value represents the time of flight (TOF) of a given pulse in the light waveform. N is an integer equal to the total number of light pulses emitted in 704 (e.g., 100 pulses). The timestamp values can then be processed by the LiDAR system and / or transmitted from the LiDAR system to an external computing device.
[0084] In 710, the computing device (e.g., Figure 1 Computing device 110 Figure 2 Onboard computing device 220 and / or Figure 4 The computing device 400 generates a counting histogram of timestamp values. The counting histogram may include plotting a one-dimensional graph of N timestamp values. Figure 10 Illustrative count histogram 1000 is provided.
[0085] In 712, the test histogram is inferred from the count histogram. The test histogram represents the number of times the photodetector of the LiDAR system is available during the detection of N pulses of the received waveform. Figure 10 The histogram shown illustrates the concepts of availability and trial. Figure 10 Fifteen different range gates are shown. At each range gate, a single pulse is emitted and a single photon is captured to detect an event (avalanche event). After the pulse is emitted, a detector is equipped at the beginning of the range gate. Thus, the detector is considered "available" for detection until an avalanche event is detected. The detector is then deactivated until the end of the range gate. During the deactivated period, the detector is unavailable for detection. Trial measurements track how many times each bar is available for detection. In the first range gate, the event is captured at bar index 2. Therefore, the detector is available for bars 0 through 2. In the second range gate, the event is captured at bar index 3. The detector is available from bars 0 through 3. Similarly, in the last range gate (15), the event is captured at bar index 1. Therefore, bars 0 through 1 are available. Avalanche event bar indices are aggregated from all integrated range gates (15 gates) to generate a count histogram. Moreover, all trial records are aggregated for each range gate to generate a trial histogram. Conceptually, the trial histogram can be inferred from the count histogram, where the first bar contains the maximum number of pulses to be integrated (15), and then for each subsequent bar, the number of trials is adjusted based on the avalanche history: Trials[j] = Max_trials - cumsum(avalanches[0:j-1]). The trials at bar j are equal to the maximum number of trials minus the sum of all avalanches from bar 0 to bar j-1.
[0086] In 714, possible intensity values for a given pulse of a received waveform are generated using a count histogram and a trial histogram. The returned intensity of the pulse depends on the reflectivity of the object. Therefore, this solution involves calculating multiple possible intensity values λ given a known pulse shape. j , where j is a predefined integer (e.g., 1-20). Each intensity value λ j It is generated based on the following mathematical equation (1), which realizes a linear combination of the counting histogram and the test histogram.
[0087]
[0088] Where s represents the span size of the entire counting histogram (e.g., the window size of the counting histogram is 100 ms) or the span size of a given portion of the counting histogram (e.g., 10 ms), f(i, τ = j) represents the known pulse shape of the waveform at bar index i, which has a centroid at waveform offset τ (i.e., the numerical sequence representing the pulse shape plus one), and ni Let k represent the test count at bar i, where i represents the bar index in the histogram corresponding to the pulse, and k i This represents the count of a given pulse detection event for the optical waveform at bar index i.
[0089] In 716, the probability fraction L is determined using the possible intensity value and the possible timestamp value. j The probability score L is determined by comparing the count histogram with each of several reference histograms given certain conditions. j . Figure 11 The illustrative reference histogram 1100 is provided in the text. Figure 11 The role of likelihood scores and reference histogram suggestions is explained. More specifically, Figure 11 The counting histograms obtained when the received intensity is 10, 20, and 30 photons per pulse are shown. The detector's sensitivity is 20% compared to the detector used in Figure 12. The top of the figure shows the likelihood values associated with each reference histogram proposal (shown in the middle section of the figure). In the middle section, three proposal histograms (minimum offset, optimal offset, and maximum offset proposal) are shown. It is shown how the optimal offset remains almost at the same bar position (bar index 5) regardless of the changing received intensity conditions. This demonstrates the effectiveness of the algorithm for the distance walk effect in the counting histogram.
[0090] Each comparison operation can be implemented using the following mathematical equation (2). Mathematical equation (2) determines the probability fraction L. j The probability score represents the timestamp value of the measurement, which is determined by the parameter τ. j and λ j The probability caused by a given possible value.
[0091]
[0092] Where λ represents the previously calculated intensity value λ. j And τ represents a timestamp value, which is set to the given previously calculated intensity value λ. j The associated predefined value j. Multiple possible values of j and intensity λ are used, respectively, for the timestamp τ. j Multiple calculated values are used to perform multiple iterations of mathematical equation (2). Each iteration provides a set of numbers, including probability scores, intensity values, and timestamp values. Illustrative sets of numbers are provided below.
[0093] Probability score L1, intensity λ1, timestamp τ = 1
[0094] Probability score L2, intensity λ2, timestamp τ = 2
[0095] …
[0096] probability fraction L 20 Intensity λ 20 Timestamp τ = 20
[0097] In 718, from the probability fractions L1, L2, ..., L... j Select possible intensity values and possible timestamp values from the possible intensity values and possible timestamp values. For example, select from intensity values λ1, λ2, ..., λ 20 Select an intensity value, and choose a timestamp value from timestamp values τ = 1, 2, ..., 20. The selected value is associated with the highest probability score (e.g., the probability score with the highest value). For example, probability score L2 is probability scores L1, L2, ..., L... 20 The maximum probability score is determined by the intensity value λ2 and the timestamp value τ = 2. This solution is not limited to the details of this example.
[0098] In 720, the estimated ranging distance d from the LiDAR system to the object is set to be equal to the selected timestamp value τ. j The estimated distance d can be defined by the following mathematical equation (3).
[0099]
[0100] For example, when selecting a timestamp value τ = 2 in 718, the estimated distance d is set to equal to 2. This solution is not limited to the details of this example.
[0101] In 722, the computing device generates the position using the estimated ranging distance d. The position can also be generated based on the LiDAR system's location, orientation, and / or angle. This position is defined according to the x, y, and z coordinates. The x-coordinate can be generated according to the following mathematical equation (4). The y-coordinate can be generated according to the following mathematical equation (5). The z-coordinate can be generated according to the following mathematical equation (6).
[0102] x=d*sin(ev)*cos(az) (4)
[0103] y=d*sin(ev)*sin(az) (5)
[0104] z = d * cos(az) (6)
[0105] Where d is the measured distance in meters, az is the horizontal angle (azimuth) with respect to the X-axis, and ev is the vertical angle (elevation) with respect to the Z-axis.
[0106] The operations described above in boxes 714-724 are repeated to generate the position and intensity values for each pulse in the received waveform, as shown in 726. Then, in 728, the position and intensity values are used to generate a LiDAR dataset. The LiDAR dataset comprises multiple data points defining a point cloud. Each data point is defined by at least a corresponding position and a corresponding intensity value. Subsequently, 730 is executed, where method 700 terminates or other operations are performed. These other operations may include, but are not limited to, controlling a vehicle (e.g., Figure 1 Operations of the AV1021 (e.g., for feature detection / extraction, LiDAR point classification (e.g., ground, vegetation, buildings, water, railways, roads, power lines, bridges, vehicles, etc.), object trajectory generation, and / or vehicle trajectory generation).
[0107] Although this solution has been shown and described with respect to one or more embodiments, equivalent changes and modifications will occur to those skilled in the art upon reading and understanding this specification and the accompanying drawings. Furthermore, while specific features of this solution may have been disclosed only with respect to one of several embodiments, such features may be combined with one or more other features of other embodiments, as may be desirable and advantageous for any given or particular application. Therefore, the breadth and scope of this solution should not be limited to any of the embodiments described above. Rather, the scope of this solution should be defined by the following claims and their equivalents.
Claims
1. A method for operating a LiDAR system, comprising: The LiDAR system receives a waveform representing the light reflected from the surface of at least one object; The LiDAR system generates multiple timestamp values for photon detection events triggered by pulses in the waveform; The processor generates a count histogram of the timestamp values; The processor infers a test histogram from the count histogram, the test histogram representing the number of times the photodetector of the LiDAR system is available during the reception of the waveform; Based on the analysis of the count histogram and the test histogram, an estimated ranging distance from the LiDAR system to the at least one object and an estimated intensity value of a given pulse of the waveform are generated, wherein the estimated intensity value is set to a maximum likelihood estimate of the pulse intensity equal to the given pulse shape, the count histogram, and the test histogram, and the maximum likelihood estimate of the pulse intensity is selected from a plurality of possible intensity values, each possible intensity value being determined using a linear combination of the count histogram and the test histogram; The processor determines the location using the estimated ranging distance from the LiDAR system to the at least one object; and The processor generates a LiDAR dataset, which includes data points defined by the location and the estimated intensity value.
2. The method of claim 1, wherein, The estimated intensity value represents an estimate of the pulse's return intensity.
3. The method of claim 1, wherein, Each of the multiple intensity values is determined according to the following mathematical equation: (1) in s This indicates the span size of at least a portion of the counting histogram. f(i, τ=j) Indicates in the bar index i The known pulse shape of the waveform at the location, the waveform at waveform offset τ It has a center of mass. n i Indicates in the vertical bar i Test count at the location, i This indicates the position of the bar index in the test histogram corresponding to the given pulse, and k i Indicates in the bar index i The number of events detected by a given pulse at a given location.
4. The method according to claim 1, wherein, The estimated distance is set to the maximum likelihood estimate of the distance given multiple possible intensity values.
5. The method according to claim 4, wherein, The maximum likelihood estimate of the ranging distance is determined by performing a comparison operation in which the counting histogram is compared with a plurality of calculated reference histograms given specific conditions.
6. The method according to claim 5, wherein, Each operation in the comparison operation generates a probability score, which represents the timestamp value determined by parameters. τ j and λ j The probability caused by a given possible value.
7. The method according to claim 6, wherein, The probability fraction is defined by the following mathematical equation: in s This indicates the span size of at least a portion of the counting histogram. f(i, τ=j) Indicates time index i The known pulse shape of the waveform at the time index. τ It has a center of mass. n i Indicates in the vertical bar i Test count at the location, i This indicates the position of the timestamp value in the test histogram corresponding to a given pulse. k i This represents the count of events detected by a given pulse. λ Represents a previously calculated strength value. λ j ,and τ This represents a timestamp value, which is set to match a previously calculated intensity value. λ j Associated predefined values j .
8. The method of claim 1, further comprising using the LiDAR dataset to control the operation of an autonomous vehicle.
9. A system comprising: processor; A non-transitory computer-readable storage medium includes programming instructions configured to cause the processor to implement a method for operating a LiDAR system, wherein the programming instructions include instructions for: Generate multiple timestamp values for a photon detection event, the photon detection event being triggered by a pulse in a received waveform representing light reflected from the surface of at least one object; Generate a count histogram of the timestamp values; The test histogram is inferred from the count histogram, which represents the number of times the photodetector of the LiDAR system is available during the reception of the waveform; Based on the analysis of the count histogram and the test histogram, an estimated ranging distance from the LiDAR system to the at least one object and an estimated intensity value of a given pulse of the received waveform are generated, wherein the estimated intensity value is set to be equal to the maximum likelihood estimate of the pulse intensity of the given pulse shape, the count histogram, and the test histogram, and the maximum likelihood estimate of the pulse intensity is selected from a plurality of possible intensity values, each possible intensity value being determined using a linear combination of the count histogram and the test histogram; The location is determined using the estimated ranging distance from the LiDAR system to the at least one object; and A LiDAR dataset is generated, which includes data points defined by the location and the estimated intensity value.
10. The system according to claim 9, wherein, The estimated intensity value represents an estimate of the pulse's return intensity.
11. The system according to claim 9, wherein, Each of the multiple intensity values is determined according to the following mathematical equation: (1) in s This indicates the span size of at least a portion of the counting histogram. f(i, τ=j) Indicates in the bar index i The known pulse shape of the waveform at the location, the waveform at waveform offset τ It has a center of mass. n i Indicates in the vertical bar i Test count at the location, i This indicates the position of the bar index in the test histogram corresponding to the given pulse, and k i Indicates in the bar index i The number of events detected by a given pulse at a given location.
12. The system according to claim 9, wherein, The estimated distance is set to the maximum likelihood estimate of the distance given multiple possible intensity values.
13. The system according to claim 12, wherein, The maximum likelihood estimate of the ranging distance is determined by performing a comparison operation in which the counting histogram is compared with a plurality of calculated reference histograms given specific conditions.
14. The system according to claim 13, wherein, Each operation in the comparison operation generates a probability score, which represents the timestamp value determined by parameters. τ j and λ j The probability caused by a given possible value.
15. The system according to claim 14, wherein, The probability fraction is defined by the following mathematical equation: in s This indicates the span size of at least a portion of the counting histogram. f(i, τ=j) Indicates time index i The known pulse shape of the waveform at the time index. τ It has a center of mass. n i Indicates in the vertical bar i Test count at the location, i This indicates the position of the timestamp value in the test histogram corresponding to a given pulse. k i This represents the count of events detected by a given pulse. λ Represents a previously calculated strength value. λ j ,and τ This represents a timestamp value, which is set to match a previously calculated intensity value. λ j Associated predefined values j .
16. The system of claim 9 further includes using the LiDAR dataset to control the operation of an autonomous vehicle.
17. A non-transitory computer-readable medium storing instructions configured to, when executed by at least one computing device, cause the at least one computing device to perform operations including: To enable the LiDAR system to receive a waveform representing light reflected from the surface of at least one object; The LiDAR system generates multiple timestamp values for photon detection events triggered by pulses in the waveform. Generate a count histogram of the timestamp values; The test histogram is inferred from the count histogram, which represents the number of times the photodetector of the LiDAR system is available during the reception of the waveform; Based on the analysis of the count histogram and the test histogram, an estimated ranging distance from the LiDAR system to the at least one object and an estimated intensity value of a given pulse of the waveform are generated, wherein the estimated intensity value is set to a maximum likelihood estimate of the pulse intensity equal to the given pulse shape, the count histogram, and the test histogram, and the maximum likelihood estimate of the pulse intensity is selected from a plurality of possible intensity values, each possible intensity value being determined using a linear combination of the count histogram and the test histogram; The location is obtained using the estimated ranging distance from the LiDAR system to the at least one object; and A LiDAR dataset is generated, which includes data points defined by the location and the estimated intensity value.
Citation Information
Patent Citations
Systems, methods, and media for single photon depth imaging with improved precision in ambient light
US20200284907A1
