Method and apparatus for pose correction of a lidar

Abstract

A method and device for correcting the posture of a laser radar. The method comprises detecting ground points in a point cloud (S202), establishing a ground coordinate system according to the ground points (S203), and calculating a posture correction parameter between a current posture of the laser radar in a radar coordinate system and a target posture in the ground coordinate system (S204), controlling a bearing device of the laser radar to rotate and / or translate based on the posture correction parameter, and adjusting the laser radar from the current posture to the target posture (S205), thereby automatically correcting the posture of the laser radar and improving the efficiency and accuracy of the posture correction of the laser radar.

Description

[0001] This application is a divisional application of Chinese application No. 202080005491.2, the foregoing contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of autonomous driving, and in particular to a method and apparatus for attitude correction of a lidar. Background Technology

[0003] In related technologies, camera pose refers to the camera's position and orientation in three-dimensional space. Camera pose, combined with the camera's viewing angle and visible distance, determines the precise range the camera can perceive. In security and autonomous driving fields, the accuracy of camera pose directly affects the performance of related functions and pedestrian safety in specific scenarios. Image-based cameras can obtain accurate pose parameters using methods such as Zhang's calibration, based on predefined texture information from a calibration board and the camera's imaging principle. However, in the pose correction process of LiDAR, because LiDAR measures spatial position rather than texture information, pose correction presents certain difficulties. How to perform pose correction on LiDAR is a problem that urgently needs to be solved. Summary of the Invention

[0004] The technical problem to be solved by the embodiments of this application is to provide a method, device and system for attitude correction of lidar, which can estimate the current attitude of lidar based on point cloud and correct the current attitude to the target attitude to improve scanning efficiency.

[0005] In a first aspect, this application provides a method for attitude correction of a lidar, comprising:

[0006] Acquire point clouds generated by LiDAR scanning; there may be one or more point clouds, and when there are multiple point clouds, each point cloud corresponds to a LiDAR scanning frame;

[0007] Acquire point clouds generated by lidar scanning;

[0008] The ground point set is obtained by detecting ground points in the point cloud;

[0009] Establish a ground coordinate system based on the aforementioned set of ground points;

[0010] The attitude correction parameters are calculated based on the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system.

[0011] The bearing device is controlled to adjust the lidar from the current attitude to the target attitude according to the attitude correction parameters.

[0012] In one possible design, detecting ground points in the point cloud to obtain a set of ground points includes:

[0013] The point cloud is divided into multiple sectors according to a preset angular interval; wherein the number of sectors is equal to 2π / △θ, and △θ is the angular interval;

[0014] The sector is divided into multiple containers according to a preset distance interval;

[0015] Determine at least one representative point among the plurality of containers;

[0016] Perform linear fitting on at least one representative point selected from each of the plurality of containers;

[0017] When the fitted straight line satisfies the slope threshold condition, at least one representative point selected from each container that satisfies the height threshold condition is taken as the ground point.

[0018] In one possible design, the ground points in the detected point cloud are used to obtain a set of ground points, including:

[0019] Obtain a pre-trained deep learning model;

[0020] The ground point set is obtained by detecting ground points in the point cloud based on the deep learning network.

[0021] In one possible design, the point cloud is a first point cloud; wherein, the step of detecting ground points in the point cloud to obtain a set of ground points includes:

[0022] When the angle between the horizontal plane of the lidar and the ground is greater than a preset angle, the normal vector of the first point cloud is determined;

[0023] The carrier device controls the lidar to scan along a first direction parallel to the ground to generate a second point cloud;

[0024] The support device controls the lidar to scan along a second direction parallel to the ground to generate a third point cloud; wherein the first direction and the second direction are perpendicular to each other.

[0025] Points in the first point cloud that are parallel to the first direction and the second direction are detected as ground points.

[0026] In one possible design, establishing a ground coordinate system based on the set of ground points includes:

[0027] Obtain a pre-trained deep learning network;

[0028] The normal vector of the ground point set is calculated based on the deep learning network;

[0029] The ground coordinate system is established based on the normal vector; wherein the x-axis and y-axis of the ground coordinate system constitute the ground.

[0030] Secondly, this application provides an attitude correction device for a lidar, comprising:

[0031] The acquisition unit is used to acquire the point cloud generated by the lidar scan;

[0032] The detection unit is used to detect ground points in the point cloud to obtain a set of ground points;

[0033] A calculation unit is used to establish a ground coordinate system based on the set of ground points;

[0034] The attitude correction parameters are calculated based on the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system.

[0035] An adjustment unit is used to control the carrier device to adjust the lidar from the current attitude to the target attitude according to the attitude correction parameters. Another aspect of this application discloses a distance compensation device based on an attitude correction device, comprising: a receiver, a transmitter, a memory, and a processor; wherein, the memory stores a set of program code, and the processor is used to call the program code stored in the memory to execute the attitude correction method of the lidar described in the above aspects.

[0036] Based on the same application concept, since the principle of the device in solving the problem and its beneficial effects can be found in the above-mentioned possible methods and implementations of distance compensation devices and their beneficial effects, the implementation of the device can be found in the implementation of the method, and repeated details will not be repeated.

[0037] Another aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described in the above aspects.

[0038] Another aspect of this application provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the methods described in the above aspects.

[0039] In this embodiment, ground points in the point cloud are detected, a ground coordinate system is established based on the ground points, and attitude correction parameters between the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system are calculated. Based on the attitude correction parameters, the lidar carrier device is controlled to rotate and / or translate, so that the lidar is adjusted from the current attitude to the target attitude, thereby realizing automatic attitude correction of the lidar and solving the problems of low efficiency and inaccuracy caused by manual attitude correction. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the architecture of the attitude correction system provided in the embodiments of this application;

[0042] Figure 2 This is a schematic flowchart of a lidar attitude correction method provided in an embodiment of this application;

[0043] Figures 3-7 This is a schematic diagram illustrating the principle of attitude correction provided in this embodiment;

[0044] Figure 8 This is a schematic diagram of a structure based on an attitude correction device provided in an embodiment of this application;

[0045] Figure 9 This is another structural schematic diagram of an attitude correction device provided in an embodiment of this application. Detailed Implementation

[0046] To make the inventive objectives, features, and advantages of the embodiments of this application more apparent and understandable, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0047] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0048] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0049] See Figure 1This is a schematic diagram of the attitude correction system provided in an embodiment of this application. The attitude correction system includes: a lidar, an attitude correction device, and a support device. The support device is used to support the lidar and includes, but is not limited to, devices such as drones, vehicles, or robotic arms. The support device can adjust the attitude of the lidar by rotation and / or translation. Figure 1 As shown, the carrier device is a drone. The drone can drive the lidar to translate along the x, y, and z axes, and rotate around the x, y, and z axes, meaning the drone can perform six degrees of freedom attitude adjustment for the lidar. An attitude correction device can also be mounted on the carrier device. This device calculates attitude correction parameters and instructs the carrier device to adjust its attitude according to these parameters, so that the lidar's current attitude is adjusted to the target attitude. The lidar emits a detection laser signal. When this signal encounters an obstacle, it reflects back to generate an echo laser signal. Obstacles include both ground and non-ground obstacles. The lidar generates a point cloud based on parameters such as the intensity of the echo signal and the distance from the obstacle to the lidar. The point cloud includes both ground and non-ground points.

[0050] Please see Figure 2 , Figure 2 This application provides an attitude correction method, which includes, but is not limited to, the following steps:

[0051] S201. Obtain the point cloud generated by the lidar scan.

[0052] The lidar can perform periodic scans, generating a point cloud (also called a data frame) after each scan. The attitude correction device can acquire one or more point clouds to be processed, meaning it can perform attitude correction on the lidar based on one or more data frames. In this embodiment, the point cloud can be a 3D point cloud, meaning it includes three-dimensional spatial coordinates and echo intensity parameters.

[0053] S202. Detect the ground points in the point cloud to obtain a set of ground points.

[0054] The points in the point cloud can be divided into two types: ground points and non-ground points. Ground points are generated by LiDAR scanning the ground, while non-ground points are generated by LiDAR scanning non-ground obstacles. In this embodiment, ground points in the point cloud can be detected by geometric analysis or machine learning algorithms. After traversing all ground points in the point cloud, a set of ground points is obtained.

[0055] In one possible implementation, the process of detecting ground points in a point cloud includes:

[0056] The point cloud is divided into multiple sectors according to a preset angular interval;

[0057] The sector is divided into multiple containers according to a preset distance interval;

[0058] Determine the representative points in each container, and perform linear fitting on the representative points in each container within the same sector to obtain the fitted line;

[0059] When the fitted line satisfies the slope threshold condition, the points in the container that satisfy the height threshold condition in the line fitting are marked as ground points.

[0060] For example: see Figure 3 As shown, the point cloud is fitted into a circle with radius r based on its distribution range. The preset angular interval is Δθ radians. The number of sectors into which the point cloud is divided is 2π / Δθ, and the sectors are: P1, P2, ..., P... 2π / Δθ For each sector, the sector is divided into multiple containers according to a preset distance interval. For example, for sector P1, sector P1 is divided into C containers: a1, a2, ..., a C C is an integer greater than 1. See also Figure 4 As shown, representative points are determined in each container. For example, the representative point is the lowest point in the container, i.e., the point with the smallest height. For sector P1, a straight line is fitted to the representative points in each container within the same sector to obtain a fitted line. Then, the slope of the fitted line is calculated. If the slope of the fitted line is less than a preset slope threshold, the points in the containers participating in the line fitting that are less than the preset height threshold are marked as ground points. If the slope of the fitted line is greater than or equal to the slope threshold, the fitting process stops. Ground points in each sector are detected using the above method.

[0061] In this embodiment, the method for detecting ground points in the point cloud is as follows:

[0062] Example A: Determining a pre-trained deep learning network;

[0063] The deep learning network detects ground points in the point cloud.

[0064] In the training phase, training samples are generated. These samples are labeled points, with the labels indicating whether the point is a ground point or a non-ground point. The deep learning network can identify whether the sample data is a ground point or a non-ground point based on the labels. Upon completion of the training phase, the trained deep learning network is used to perform a testing phase. In the testing phase, the points in the point cloud generated in S201 are input into the trained deep learning network to detect whether the point type is a ground point or a non-ground point. In this embodiment, the deep learning network can be a PointNet++ network.

[0065] Example B: The process of detecting ground points in a point cloud includes:

[0066] When the angle between the horizontal plane of the lidar and the ground is greater than a preset angle, the normal vector of the first point cloud is determined;

[0067] The lidar is controlled by the carrier device to scan and generate a second point cloud along a first direction parallel to the ground. It should be noted that if the carrier device is a vehicle, the vehicle's direction of travel remains parallel to the ground, and it is only necessary to control the vehicle to travel in different directions; if the carrier device is a drone, the drone needs to be controlled to fly in a direction parallel to the ground.

[0068] The lidar is controlled by the supporting device to scan along a second direction parallel to the ground to generate a third point cloud; wherein the first direction and the second direction are perpendicular to each other. Optionally, the first direction and the second direction may not be perpendicular to each other. Sometimes it is not easy to control the lidar's supporting device to move along the ground direction. After finding the normal vector of the ground coordinate system, the direction of movement at a certain moment is the x-direction. The third direction of the ground coordinate system can be obtained by cross product of the normal vector and the x-direction.

[0069] Points in the first point cloud that are parallel to the first direction and the second direction are detected as ground points.

[0070] In this embodiment, Examples A and B are applicable to ground point detection in a single frame. When the angle between the horizontal plane of the lidar and the ground is greater than a preset angle, for example, a preset angle of 20°, to improve the robustness of ground point detection, this embodiment can detect ground points based on multiple frames: at time t1, the point cloud p1 generated by the lidar scan is acquired, and the normal vector of the point cloud p1 is calculated based on K-neighborhood or PCA (Principal Component Analysis); at time t2, the attitude correction device controls the lidar to move a certain distance (e.g., 5m) along a first direction parallel to the ground through the support device. After the lidar scan, the point cloud p2 is obtained. The attitude correction device uses the ICP (Iterative Closest Point) algorithm or NDT (Normal Distributions) algorithm to detect the ground points. The Transform (Normal Distribution Transform) algorithm estimates the first motion trajectory of the lidar, obtaining the x-axis of the ground coordinate system. The attitude correction device, controlled by the carrier, moves the lidar a certain distance (e.g., 5m) along a second direction parallel to the ground. After scanning, the lidar obtains point cloud p3. The first and second directions are perpendicular to each other. The attitude correction device estimates the second motion trajectory of the lidar according to the ICP or NDT algorithm, thus obtaining the y-axis of the ground coordinate system. The normal vectors of the x-axis and y-axis are the normal vectors of the ground points, thereby detecting the ground points in point cloud p1.

[0071] It should be noted that the specific value of the preset angle may be related to the ground point detection method. For example, when the preset angle is 5°, the ground point detection method in Example A is no longer applicable; when the preset angle is 8°, the ground point detection algorithm in Example B is no longer applicable. Furthermore, the specific value of the preset angle may also be related to the performance of the LiDAR. For example, the number of scan lines of the LiDAR is positively correlated with the preset angle; the more scan lines, the larger the preset angle value, and vice versa.

[0072] S203. Establish a ground coordinate system based on the set of ground points.

[0073] In this system, the coordinates and orientation of the ground coordinate system represent the attitude of the ground based on the current radar coordinate system or the radar coordinate system. This attitude is used to establish the ground coordinate system. In the ground coordinate system, the z-axis is perpendicular to the ground, and the plane formed by the x-axis and y-axis is the ground surface. This embodiment can establish the ground coordinate system using the PCA algorithm or a deep learning network. For example, the established ground coordinate system is as follows: Figure 6 As shown.

[0074] In one possible implementation, see Figure 5 As shown, Figure 5 Given the set of ground points projected onto the x-axis and y-axis, determine the two perpendicular first direction vectors and second direction vectors in the set according to PCA. The first direction vector and the second direction vector are the longest and widest directions in the set, respectively. Then determine the normal vector perpendicular to the x-axis and y-axis, which is the z-axis.

[0075] In one possible implementation, a ground coordinate system is established based on a deep learning network. During the training phase, training samples are generated; these samples are labeled points, with each label representing the point's normal vector. The deep learning network can identify the normal vector of the sample data based on the labels. Upon completion of the training phase, a testing phase is performed using the trained deep learning network. During the testing phase, the points in the ground point set generated in S202 are input into the trained deep learning network to identify the normal vectors. In this embodiment, the deep learning network can be a PointNet++ network.

[0076] S204. Calculate attitude correction parameters based on the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system.

[0077] In this system, the lidar is located at the origin of the radar coordinate system. The plane formed by the x-axis and y-axis of the radar coordinate system is the horizontal plane of the lidar, and the z-axis is perpendicular to this horizontal plane. For example, see the radar coordinate system diagram. Figure 7As shown. The attitude correction device is pre-configured with the target attitude based on the ground coordinate system. The attitude correction device can calculate the attitude correction parameters between the current attitude in the radar coordinate system and the target attitude in the ground coordinate system according to the spatial geometric relationship. The attitude correction parameters include rotation (rx, ry, rz) and translation (dx, dy, dz). The rotation represents the angle of rotation around the x-axis, y-axis or z-axis, and the translation is...

[0078] In this system, the radar coordinate system is used as the original coordinate system. Two objects, O1 and O2, are located at two positions and facing two directions within this coordinate system. The attitude of O1 is T1 (a matrix representing O1's spatial position and orientation in the original coordinate system, which can be decomposed into T1 = Mr1 * Mt1, representing the object's position offset Mt1 and rotation offset Mr1 relative to the original coordinate system). The attitude of O2 is T2 (similarly, T1 represents O2's other spatial position and orientation in the original coordinate system, and can also be expressed as T2 = Mr2 * Mt2, representing the object's position offset Mt2 and rotation offset Mr2 relative to the original coordinate system). Assume there is a transformation matrix M (a matrix of [dx, dy, dz, rx, ry, rz]) such that T1 can coincide with T2 through M. Mathematically, this is expressed as T2 = M * T1. Then, M = Inv(Mr1) * Mr2 * (Mt2 - Mt1). Here, Inv(·) represents the matrix inversion operation. T1 can be considered as the current attitude of the radar (by default, it coincides with the radar coordinate system, i.e., T1 is the identity matrix), T2 can be considered as the attitude of the ground or the given target attitude, and M is the attitude correction parameter.

[0079] For example, attitude correction parameters can be represented using a transformation matrix, which represents the rotation and translation amounts to make the radar coordinate system and the ground coordinate system coincide. Assume the radar coordinate system origin is (0, 0, 0), the radar coordinate system orientation is (0, 0, 1) along the z-axis, (1, 0, 0) along the x-axis, and (0, 1, 0) along the y-axis. The ground coordinate system origin is (1, 0, 0), the z-axis / normal vector is (0, 0, 1), the x-axis is (0, 1, 0), and the y-axis is (-1, 0, 0). Then, the rotation and translation amounts are: translate the origin of the radar coordinate system by 1 unit along the x-axis, and then rotate the x-axis and y-axis counterclockwise by 90 degrees to make the radar coordinate system coincide with the ground coordinate system.

[0080] S205. Control the carrier device to adjust the lidar from the current attitude to the target attitude according to the attitude correction parameters.

[0081] The supporting device can be a six-degree-of-freedom movable device. The attitude correction device drives the supporting device to adjust its attitude according to the attitude correction parameters calculated in S204, so that the lidar can be adjusted from its current attitude to a template attitude. See also Figure 7As shown, the calibration device adjusts the attitude of the lidar according to the attitude calibration parameters so that the horizontal plane of the lidar is parallel to the ground.

[0082] according to Figure 2 The description describes the process of detecting ground points in a point cloud, establishing a ground coordinate system based on these points, and calculating attitude correction parameters between the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system. Based on these attitude correction parameters, the lidar's carrier device is controlled to rotate and / or translate, adjusting the lidar from its current attitude to the target attitude. This achieves automatic attitude correction of the lidar, solving the problems of low efficiency and inaccuracy caused by manual attitude correction.

[0083] The above describes in detail a method for attitude correction of a lidar according to an embodiment of this application. Below, an attitude correction device (hereinafter referred to as device 3) according to an embodiment of this application is provided.

[0084] Figure 8 The device 3 shown can achieve Figure 2 The attitude correction method for lidar in the embodiment shown includes an acquisition unit 301, a detection unit 302, a calculation unit 303, and an adjustment unit 304.

[0085] Acquisition unit 301 is used to acquire point clouds generated by lidar scanning;

[0086] Detection unit 302 is used to detect ground points in the point cloud to obtain a set of ground points;

[0087] Calculation unit 303 is used to establish a ground coordinate system based on the set of ground points;

[0088] The attitude correction parameters are calculated based on the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system.

[0089] The adjustment unit 304 is used to control the carrier device to adjust the lidar from the current attitude to the target attitude according to the attitude correction parameters.

[0090] Optionally, the step of detecting ground points in the point cloud to obtain a set of ground points includes:

[0091] The point cloud is divided into multiple sectors according to a preset angular interval;

[0092] The sector is divided into multiple containers according to a preset distance interval;

[0093] Determine the representative points in each container, and perform linear fitting on the representative points in each container within the same sector to obtain the fitted line.

[0094] When the fitted line satisfies the slope threshold condition, the points in the container that satisfy the height threshold condition in the line fitting are marked as ground points.

[0095] Optionally, the step of detecting ground points in the point cloud to obtain a set of ground points includes:

[0096] Obtain a pre-trained deep learning network;

[0097] The ground point set is obtained by detecting ground points in the point cloud using the deep learning network.

[0098] Optionally, the angle between the horizontal plane of the lidar and the ground is less than or equal to a preset angle.

[0099] Optionally, the point cloud is a first point cloud;

[0100] The step of detecting ground points in the point cloud to obtain a set of ground points includes:

[0101] When the angle between the horizontal plane of the lidar and the ground is greater than a preset angle, the normal vector of the first point cloud is determined;

[0102] The carrier device controls the lidar to scan along a first direction parallel to the ground to generate a second point cloud;

[0103] The support device controls the lidar to scan along a second direction parallel to the ground to generate a third point cloud; wherein the first direction and the second direction are perpendicular to each other.

[0104] Points in the first point cloud that are parallel to the first direction and the second direction are detected as ground points.

[0105] Optionally, establishing a ground coordinate system based on the set of ground points includes:

[0106] Obtain a pre-trained deep learning network;

[0107] The normal vector of the ground point set is calculated based on the deep learning network;

[0108] The ground coordinate system is established based on the normal vector; wherein the x-axis and y-axis of the ground coordinate system constitute the ground.

[0109] This application's embodiments and Figures 1 to 7 The method implementations are based on the same concept and bring about the same technical effects. The specific process can be referred to Figures 1 to 7 The method embodiments described herein will not be repeated here.

[0110] The device 3 can be a field-programmable gate array (FPGA), a dedicated integrated chip, a system on chip (SoC), a central processing unit (CPU), a network processor (NP), a digital signal processing circuit, a micro controller unit (MCU), or a programmable logic device (PLD) or other integrated chips to implement the relevant functions.

[0111] The above describes in detail a lidar attitude correction method according to an embodiment of this application. Below, an attitude correction device (hereinafter referred to as device 4) according to an embodiment of this application is provided.

[0112] Figure 9 This is a schematic diagram of a device structure provided in an embodiment of this application, hereinafter referred to as device 4. Device 4 can be integrated into the lidar or carrier device of the above embodiments, such as... Figure 4 As shown, the device includes: a memory 402, a processor 401, a transmitter 404, and a receiver 403.

[0113] The memory 402 can be a separate physical unit, which can be connected to the processor 401, transmitter 404, and receiver 403 via a bus. The memory 402, processor 401, transmitter 404, and receiver 401 can also be integrated together, implemented in hardware, etc.

[0114] Transmitter 404 is used to transmit signals, and receiver 403 is used to receive signals.

[0115] The memory 402 is used to store programs that implement the above method embodiments or the various modules of the device embodiments. The processor 401 calls the program to execute the operations of the above method embodiments.

[0116] Optionally, when some or all of the attitude correction method of the lidar in the above embodiments is implemented by software, the device may also include only a processor. The memory for storing the program is located outside the device, and the processor is connected to the memory via circuits / wires to read and execute the program stored in the memory.

[0117] The processor can be a central processing unit (CPU), a network processor (NP), or a combination of a CPU and an NP.

[0118] The processor may further include hardware chips. These hardware chips may be application-specific integrated circuits (ASICs), programmable logic devices (PLDs), or combinations thereof. The PLDs may be complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), generic array logic (GALs), or any combination thereof.

[0119] Memory may include volatile memory, such as random-access memory (RAM); memory may also include non-volatile memory, such as flash memory, hard disk drive (HDD), or solid-state drive (SSD); memory may also include combinations of the above types of memory.

[0120] In the above embodiments, the transmitting unit or transmitter performs the transmitting steps of the various method embodiments described above, and the receiving unit or receiver performs the receiving steps of the various method embodiments described above. Other steps are performed by other units or processors. The transmitting unit and the receiving unit can form a transceiver unit, and the receiver and the transmitter can form a transceiver.

[0121] This application also provides a computer storage medium storing a computer program for executing the attitude correction method for lidar provided in the above embodiments.

[0122] This application also provides a computer program product containing instructions that, when run on a computer, causes the computer to execute the attitude correction method for the lidar provided in the above embodiments.

[0123] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0124] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0125] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0126] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

Claims

1. An attitude correction device, characterized in that, include: The acquisition unit is used to acquire the point cloud generated by the lidar scan; The detection unit is used to detect ground points in the point cloud to obtain a set of ground points; A calculation unit is used to establish a ground coordinate system based on the set of ground points; The attitude correction parameters are calculated based on the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system. The adjustment unit is used to control the carrier device to adjust the lidar from the current attitude to the target attitude according to the attitude correction parameters; The point cloud is a first point cloud; wherein, the detection unit is specifically used for, When the angle between the horizontal plane of the lidar and the ground is greater than a preset angle, the normal vector of the first point cloud is determined; The carrier device controls the lidar to scan along a first direction parallel to the ground to generate a second point cloud; The support device controls the lidar to scan along a second direction parallel to the ground to generate a third point cloud; wherein the first direction and the second direction are perpendicular to each other. Points in the first point cloud that are parallel to the first direction and the second direction are detected as ground points.

2. The calibration device according to claim 1, characterized in that, The detection unit is specifically used to divide the point cloud into multiple sectors according to a preset angular interval; The sector is divided into multiple containers according to a preset distance interval; Determine the representative points in each container, and perform linear fitting on the representative points in each container within the same sector to obtain the fitted line. When the fitted line satisfies the slope threshold condition, the points in the container that satisfy the height threshold condition in the line fitting are marked as ground points.

3. The calibration device according to claim 1, characterized in that, The detection unit is specifically used for, Obtain a pre-trained deep learning network; The ground point set is obtained by detecting ground points in the point cloud using the deep learning network.

4. The calibration device according to claim 2 or 3, characterized in that, The angle between the horizontal plane of the lidar and the ground is less than or equal to a preset angle.

5. The calibration device according to claim 1, characterized in that, The computing unit is used for, Obtain a pre-trained deep learning network; The normal vector of the ground point set is calculated based on the deep learning network; The ground coordinate system is established based on the normal vector; wherein the x-axis and y-axis of the ground coordinate system constitute the ground.

6. A method for attitude correction of a lidar, characterized in that, include: Acquire point clouds generated by lidar scanning; The ground point set is obtained by detecting ground points in the point cloud; Establish a ground coordinate system based on the set of ground points; The attitude correction parameters are calculated based on the current attitude of the lidar in the radar coordinate system and the target attitude in the ground coordinate system. According to the attitude correction parameters, the carrier device is controlled to adjust the lidar from the current attitude to the target attitude; The point cloud is a first point cloud; wherein, the step of detecting ground points in the point cloud to obtain a set of ground points includes: When the angle between the horizontal plane of the lidar and the ground is greater than a preset angle, the normal vector of the first point cloud is determined; The carrier device controls the lidar to scan along a first direction parallel to the ground to generate a second point cloud; The support device controls the lidar to scan along a second direction parallel to the ground to generate a third point cloud; wherein the first direction and the second direction are perpendicular to each other. Points in the first point cloud that are parallel to the first direction and the second direction are detected as ground points.

Images