Calibration method and device for external parameters of laser radar installed on vehicle and vehicle

Abstract

The invention provides a calibration method and device for external parameters of a laser radar installed on a vehicle and the vehicle. The method comprises the following steps: in a vehicle driving process, acquiring continuous-time multi-frame point cloud data acquired by a laser radar; based on the multi-frame point cloud data, determining a target yaw angle of the laser radar relative to the vehicle; acquiring other external parameters of the laser radar corresponding to each frame of point cloud data according to the multiple frames of point cloud data; according to other external parameters of the laser radar corresponding to each frame of point cloud data, performing data statistics of the other external parameters to obtain at least one piece of intermediate statistical data; and performing discrete evaluation based on the at least one piece of intermediate statistical data to determine other target external parameters of the lidar. According to the method, accurate calibration of the external parameters of the vehicle-mounted laser radar can be completed when a vehicle is on a common road surface.

Description

Technical Field

[0001] The present application relates to the field of vehicles, and in particular to a method, device and vehicle for calibrating external parameters of a laser radar installed on a vehicle. Background Art

[0002] As the core sensor of autonomous vehicles, the accuracy of the external parameters of lidar relative to the vehicle body directly determines the registration accuracy of the collected point cloud data with the vehicle coordinate system.

[0003] Currently, existing LiDAR extrinsic parameter calibration methods are divided into offline calibration methods and online calibration methods. The offline calibration method requires data collection at a specific calibration site and equipment, followed by offline calculation of the LiDAR's extrinsic parameters. This results in a cumbersome calibration process and high calibration costs. The online calibration method calibrates the LiDAR's extrinsic parameters while the vehicle is in motion. However, existing online calibration methods require the vehicle to be driven under extremely idealized conditions (such as an absolutely level ground or a straight line) to ensure accurate calibration of the LiDAR's extrinsic parameters. However, in actual applications, such extremely idealized road conditions do not exist. Therefore, the LiDAR extrinsic parameters calibrated using existing online calibration methods often suffer from poor calibration accuracy.

[0004] In summary, there is an urgent need for a method that can enable vehicles to accurately calibrate the external parameters of the lidar on ordinary roads. Summary of the Invention

[0005] The embodiments of the present application provide a method, device, and vehicle for calibrating the external parameters of a laser radar installed on a vehicle, which are used to achieve the effect of accurately calibrating the external parameters of the vehicle-mounted laser radar on ordinary roads.

[0006] In a first aspect, an embodiment of the present application provides a method for calibrating external parameters of a laser radar installed on a vehicle, comprising:

[0007] During the driving of the vehicle, obtaining time-continuous multi-frame point cloud data collected by the laser radar;

[0008] Determining a target yaw angle of the laser radar relative to the vehicle based on the multi-frame point cloud data;

[0009] According to the multiple frames of point cloud data, other external parameters of the laser radar corresponding to each frame of point cloud data are obtained;

[0010] Performing data statistics of other extrinsic parameters of the laser radar corresponding to each frame of point cloud data to obtain at least one intermediate statistical data;

[0011] A discrete evaluation is performed based on the at least one intermediate statistical data to determine other external parameters of the target of the laser radar.

[0012] This method uses time-continuous multi-frame point cloud data to determine the target yaw angle, and eliminates the influence of data fluctuations on ordinary roads on the calibration accuracy by performing data statistics and discrete evaluation on other external parameters of the lidar corresponding to the multi-frame point cloud data, thereby ensuring the accuracy of the calibrated external parameters, thereby achieving the effect of accurately calibrating the external parameters of the lidar while driving on ordinary roads.

[0013] In a possible implementation, performing data statistics of other extrinsic parameters of the laser radar corresponding to each frame of point cloud data to obtain at least one intermediate statistical data includes:

[0014] Segmenting data of other extrinsic parameters of the laser radar corresponding to the multi-frame point cloud data to obtain at least one data segment corresponding to the other extrinsic parameters;

[0015] The mean of each data segment is calculated and used as the intermediate statistical data.

[0016] This embodiment first divides the data into segments and then calculates the mean of each segment, and uses the mean as a means of intermediate statistical data. While retaining the local characteristics of the overall data, it reduces the single-frame error through segmented smoothing, thereby improving the reliability and accuracy of other target external parameters determined based on the intermediate statistical data.

[0017] In a possible implementation, the performing a discrete evaluation based on the at least one intermediate statistical data to determine other target extrinsic parameters of the laser radar includes:

[0018] Performing discrete evaluation based on the at least one intermediate statistical data to obtain an evaluation result;

[0019] If the evaluation result meets the preset conditions, determining other target external parameters of the laser radar based on the at least one intermediate statistical data;

[0020] If the evaluation result does not meet the preset conditions, multiple frames of new point cloud data that are time-continuous with the multiple frames of point cloud data are obtained to obtain at least one new intermediate statistical data, and based on the at least one new intermediate statistical data and the at least one intermediate statistical data, other external parameters of the target of the lidar are determined.

[0021] This embodiment determines intermediate statistical data used to calculate other target external parameters based on whether the discrete evaluation results meet preset conditions, so as to ensure the accuracy of the determined target external parameters.

[0022] In a possible implementation, performing discrete evaluation on the at least one intermediate statistical data to obtain an evaluation result includes:

[0023] calculating a variance of the at least one intermediate statistical data, and using the variance as the evaluation result;

[0024] The preset condition is that the variance of the at least one intermediate statistic is smaller than the preset variance corresponding to the other external parameters.

[0025] This embodiment performs a discrete evaluation of the data based on the variance of the currently acquired intermediate statistical data, so as to realize the fluctuation analysis of the data through simple processing means.

[0026] In one possible implementation, acquiring multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new intermediate statistical data, and determining other target extrinsic parameters of the lidar based on the at least one new intermediate statistical data and the at least one intermediate statistical data, includes:

[0027] Acquire multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new data segment;

[0028] Calculating a mean value of each new data segment, and using the mean value of each new data segment as the new intermediate statistical data;

[0029] performing a discrete evaluation based on the at least one new intermediate statistical data and the at least one intermediate statistical data to obtain a new evaluation result;

[0030] If the new evaluation result does not meet the preset conditions, repeat the above steps until the new evaluation result meets the preset conditions, and determine other external parameters of the target of the laser radar based on the at least one intermediate statistical data and the at least one new intermediate statistical data.

[0031] This embodiment continuously acquires new point cloud data in a cyclic manner until the intermediate statistical data meets the preset conditions, so as to ensure the reliability of the intermediate statistical data used to determine other external parameters of the target. At the same time, it also eliminates the impact of uneven road surface on the accuracy of external parameter calibration, thereby ensuring the accuracy of other external parameters of the target that are finally determined.

[0032] In a possible implementation, determining other target extrinsic parameters of the laser radar based on the at least one intermediate statistical data includes:

[0033] Using the mean of the at least one intermediate statistical data as another target extrinsic parameter of the laser radar;

[0034] Accordingly, determining other target extrinsic parameters of the laser radar based on the at least one intermediate statistical data and the at least one new intermediate statistical data includes:

[0035] The at least one intermediate statistical data and the mean value of the at least one new intermediate statistical data are used as other target external parameters of the laser radar.

[0036] This embodiment calculates final target external parameters by averaging intermediate statistical data to achieve rapid calibration of the target external parameters.

[0037] In a possible implementation, segmenting data of other extrinsic parameters of the laser radar corresponding to the multiple frames of point cloud data to obtain at least one data segment corresponding to the other extrinsic parameters includes:

[0038] The data of other external parameters of the laser radar corresponding to the multi-frame point cloud data are divided into at least one data segment according to the preset change range corresponding to the other external parameters, with the data of the other external parameters exceeding the preset change range as the dividing point, and the data exceeding the preset change range in each data segment is deleted.

[0039] This embodiment eliminates data with large fluctuations in other external parameters during the data segment acquisition stage, thereby accelerating the speed of making discrete evaluation results meet preset conditions, thereby further improving the accuracy of external parameter calibration results and the calibration speed effect of external parameters.

[0040] In a possible implementation, the other external parameters include at least one of a roll angle of the laser radar relative to the vehicle, a pitch angle of the laser radar relative to the vehicle, and a height of the laser radar above the ground.

[0041] This implementation is used to provide external parameters that can be actually calibrated using this solution, providing a practical solution for the specific implementation of the solution.

[0042] In one possible implementation, determining the target yaw angle of the laser radar relative to the vehicle based on the multi-frame point cloud data includes:

[0043] Determine the coordinate system of the first frame of point cloud data in the multiple frames of point cloud data as the odometry coordinate system of the laser radar;

[0044] Obtaining, in the multiple frames of point cloud data, a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data, and a vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data;

[0045] Based on the vehicle yaw angle and the radar yaw angle, a target yaw angle of the lidar relative to the vehicle is determined.

[0046] This embodiment uses the lidar odometry as an intermediary, and directly obtains the yaw angle of the lidar relative to the vehicle at any time through the radar yaw angle and the vehicle yaw angle, so that the vehicle can drive normally on ordinary roads (can turn, no need to go straight), and the target yaw angle can be accurately calibrated.

[0047] In a possible implementation, obtaining, in the multiple frames of point cloud data, a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data, and a vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data, includes:

[0048] In the mileage coordinate system, based on the multi-frame point cloud data, obtaining a position change of each frame of the laser radar relative to the previous frame; wherein the position change includes a yaw angle change and a translation change of the laser radar;

[0049] Determining a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data according to a change in the yaw angle of each frame of the laser radar relative to the previous frame;

[0050] The vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data is determined based on the translation change of each frame of the laser radar relative to the previous frame.

[0051] This embodiment provides an implementation method for obtaining the radar yaw angle and the vehicle yaw angle based on the lidar odometer, providing a basis for determining the target yaw angle based on the radar yaw angle and the vehicle yaw angle.

[0052] In one possible implementation, determining the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data based on the translation change of each frame of the laser radar relative to the previous frame includes:

[0053] Dividing the multi-frame point cloud data into a plurality of point cloud data segments according to a preset number of frames;

[0054] In each point cloud data segment, according to the translation change of each frame of the laser radar relative to the previous frame, the coordinate data of the laser radar in the odometry coordinate system corresponding to each frame is obtained;

[0055] Performing curve fitting on the coordinate data of the laser radar corresponding to each frame in the point cloud data segment to obtain a curve equation;

[0056] For a moment when any frame of point cloud data is collected in the point cloud data segment, a derivative of the curve equation with respect to the moment is calculated to obtain a vehicle yaw angle of the vehicle in the odometer coordinate system when the frame of point cloud data is collected.

[0057] This embodiment quickly obtains the vehicle's yaw angle in the odometer coordinate system by performing curve fitting on the odometer trajectory (i.e., the coordinate data of each frame of the lidar in the odometer coordinate system) and calculating the derivative of the fitted curve equation at the moment when any frame of point cloud data is collected.

[0058] In one possible implementation, determining the target yaw angle of the lidar relative to the vehicle based on the vehicle yaw angle and the radar yaw angle includes:

[0059] For each frame of point cloud data, the difference between the vehicle yaw angle and the radar yaw angle is calculated to obtain the yaw angle of the lidar relative to the vehicle;

[0060] According to the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data, the target yaw angle of the laser radar relative to the vehicle is obtained.

[0061] In this embodiment, the yaw angle of the lidar relative to the vehicle is directly obtained through the difference between the vehicle's yaw angle and the radar's yaw angle, so that there is no need to limit the vehicle to execute it. The yaw angle of the lidar relative to the vehicle at each moment can be obtained when the vehicle is driving normally (for example, it can turn). In addition, this embodiment determines the target based on the yaw angle of the lidar relative to the vehicle corresponding to multiple frames of point cloud data, so as to eliminate the instability of a single data and realize accurate calibration of the target yaw angle.

[0062] In a possible implementation, obtaining a target yaw angle of the lidar relative to the vehicle based on the yaw angle of the lidar relative to the vehicle corresponding to each frame of point cloud data in the multiple frames of point cloud data includes:

[0063] performing data statistics of the yaw angle of the laser radar relative to the vehicle, based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data in the multiple frames of point cloud data, to obtain at least one intermediate statistical data;

[0064] A discrete evaluation is performed based on the at least one intermediate statistic to determine a target yaw angle of the lidar relative to the vehicle.

[0065] This embodiment eliminates the influence of the instability of a single data on the accuracy of yaw angle calibration by performing data statistics and discrete estimation processing on the yaw angle of the lidar relative to the vehicle corresponding to multiple frames of point cloud data, thereby achieving accurate calibration of the lidar yaw angle based on the point cloud data obtained when the vehicle is driving on ordinary roads.

[0066] In a second aspect, an embodiment of the present application provides a device for calibrating extrinsic parameters of a laser radar installed on a vehicle, comprising:

[0067] An acquisition unit, configured to acquire time-continuous multi-frame point cloud data collected by the laser radar during the driving of the vehicle;

[0068] a first processing unit, configured to determine a target yaw angle of the laser radar relative to the vehicle based on the multiple frames of point cloud data;

[0069] A second processing unit is configured to obtain other external parameters of the laser radar corresponding to each frame of point cloud data based on the multiple frames of point cloud data;

[0070] a third processing unit, configured to perform data statistics of other extrinsic parameters of the laser radar corresponding to each frame of point cloud data, to obtain at least one intermediate statistical data;

[0071] A fourth processing unit is used to perform discrete evaluation based on the at least one intermediate statistical data to determine other external parameters of the target of the laser radar.

[0072] In a possible implementation, the third processing unit includes:

[0073] A segmentation module, configured to segment data of other external parameters of the laser radar corresponding to the multi-frame point cloud data to obtain at least one data segment corresponding to the other external parameters;

[0074] The calculation module is used to calculate the mean of each data segment and use the mean of each data segment as the intermediate statistical data.

[0075] In a possible implementation, the fourth processing unit includes:

[0076] an evaluation module, configured to perform discrete evaluation based on the at least one intermediate statistical data and obtain an evaluation result;

[0077] A first determination module is configured to determine other target extrinsic parameters of the laser radar based on the at least one intermediate statistical data if the evaluation result meets a preset condition;

[0078] The second determination module is used to obtain multiple frames of new point cloud data that are time-continuous with the multiple frames of point cloud data if the evaluation result does not meet the preset conditions, so as to obtain at least one new intermediate statistical data, and determine other external parameters of the target of the lidar based on the at least one new intermediate statistical data and the at least one intermediate statistical data.

[0079] In a possible implementation, the evaluation module is specifically configured to:

[0080] calculating a variance of the at least one intermediate statistical data, and using the variance as the evaluation result;

[0081] The preset condition is that the variance of the at least one intermediate statistic is smaller than the preset variance corresponding to the other external parameters.

[0082] In a possible implementation manner, the second determining module is specifically configured to:

[0083] Acquire multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new data segment;

[0084] Calculating a mean value of each new data segment, and using the mean value of each new data segment as the new intermediate statistical data;

[0085] performing a discrete evaluation based on the at least one new intermediate statistical data and the at least one intermediate statistical data to obtain a new evaluation result;

[0086] If the new evaluation result does not meet the preset conditions, repeat the above steps until the new evaluation result meets the preset conditions, and determine other external parameters of the target of the laser radar based on the at least one intermediate statistical data and the at least one new intermediate statistical data.

[0087] In a possible implementation, the first determining module is partially configured to:

[0088] Using the mean of the at least one intermediate statistical data as another target extrinsic parameter of the laser radar;

[0089] Accordingly, the second determining module is partially used to:

[0090] The at least one intermediate statistical data and the mean value of the at least one new intermediate statistical data are used as other target external parameters of the laser radar.

[0091] In a possible implementation, the segmentation module is specifically configured to:

[0092] The data of other external parameters of the laser radar corresponding to the multi-frame point cloud data are divided into at least one data segment according to the preset change range corresponding to the other external parameters, with the data of the other external parameters exceeding the preset change range as the dividing point, and the data exceeding the preset change range in each data segment is deleted.

[0093] In a possible implementation, the other external parameters in the second processing unit include at least one of a roll angle of the laser radar relative to the vehicle, a pitch angle of the laser radar relative to the vehicle, and a height of the laser radar above the ground.

[0094] In a possible implementation, the first processing unit includes:

[0095] A third determining module is used to determine the coordinate system of the first frame of point cloud data in the multiple frames of point cloud data as the odometry coordinate system of the laser radar;

[0096] an acquisition module, configured to acquire, in the multiple frames of point cloud data, a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data, and a vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data;

[0097] A fourth determination module is used to determine a target yaw angle of the laser radar relative to the vehicle based on the vehicle yaw angle and the radar yaw angle.

[0098] In a possible implementation, the acquisition module includes:

[0099] A first submodule is configured to obtain, in the mileage coordinate system, a position change of each frame of the laser radar relative to the previous frame based on the multi-frame point cloud data; wherein the position change includes a yaw angle change and a translation change of the laser radar;

[0100] The second submodule is configured to determine the radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data based on the yaw angle change of each frame of the laser radar relative to the previous frame;

[0101] The third submodule is used to determine the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data based on the translation change of each frame of the laser radar relative to the previous frame.

[0102] In a possible implementation, the third submodule is specifically configured to:

[0103] Dividing the multi-frame point cloud data into a plurality of point cloud data segments according to a preset number of frames;

[0104] In each point cloud data segment, according to the translation change of each frame of the laser radar relative to the previous frame, the coordinate data of the laser radar in the odometry coordinate system corresponding to each frame is obtained;

[0105] Performing curve fitting on the coordinate data of the laser radar corresponding to each frame in the point cloud data segment to obtain a curve equation;

[0106] For a moment when any frame of point cloud data is collected in the point cloud data segment, a derivative of the curve equation with respect to the moment is calculated to obtain a vehicle yaw angle of the vehicle in the odometer coordinate system when the frame of point cloud data is collected.

[0107] In a possible implementation, the fourth determining module includes:

[0108] A fourth submodule is configured to calculate, for each frame of point cloud data, the difference between the vehicle yaw angle and the radar yaw angle to obtain the yaw angle of the lidar relative to the vehicle;

[0109] The fifth submodule is used to obtain the target yaw angle of the laser radar relative to the vehicle based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data.

[0110] In a possible implementation manner, the fifth submodule is specifically configured to:

[0111] performing data statistics of the yaw angle of the laser radar relative to the vehicle, based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data in the multiple frames of point cloud data, to obtain at least one intermediate statistical data;

[0112] A discrete evaluation is performed based on the at least one intermediate statistic to determine a target yaw angle of the lidar relative to the vehicle.

[0113] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory, a processor;

[0114] The memory stores computer-executable instructions;

[0115] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0116] In a fourth aspect, an embodiment of the present application provides a vehicle, comprising: a vehicle body and the electronic device described in the third aspect.

[0117] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed, they are used to implement the first aspect and / or various possible implementation methods of the first aspect as described above.

[0118] In a sixth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0119] The embodiments of the present application provide a method, device and vehicle for calibrating the external parameters of a laser radar installed on a vehicle, which obtains multi-frame point cloud data collected by the laser radar in a time-continuous manner during vehicle driving; determines the target yaw angle of the laser radar relative to the vehicle based on the multi-frame point cloud data; obtains other external parameters of the laser radar corresponding to each frame of point cloud data based on the multi-frame point cloud data; performs data statistics of the other external parameters based on the other external parameters of the laser radar corresponding to each frame of point cloud data to obtain at least one intermediate statistical data; performs discrete evaluation based on the at least one intermediate statistical data to determine the means of determining other target external parameters of the laser radar, and calculates the target yaw angle using multi-frame point cloud data. At the same time, by performing data statistics and discrete evaluation on other external parameters of the laser radar corresponding to the multi-frame point cloud data, the influence of data fluctuations on the calibration accuracy in ordinary roads is eliminated, and the accuracy of the calibrated external parameters is ensured, thereby achieving the effect of accurately calibrating the external parameters of the vehicle-mounted laser radar on ordinary roads. BRIEF DESCRIPTION OF THE DRAWINGS

[0120] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0121] Figure 1 A schematic diagram of the process of calibrating the external parameters of a vehicle-mounted laser radar provided in this application;

[0122] Figure 2 A schematic diagram of the structure of the device for calibrating the external parameters of a laser radar installed on a vehicle provided in this application;

[0123] Figure 3 This is a schematic diagram of the structure of the electronic device provided in this application.

[0124] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0125] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0126] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0127] Figure 1 The flow chart of the method for calibrating the external parameters of the laser radar installed on the vehicle provided in this application is as follows: Figure 1 As shown, the method includes:

[0128] S101. During vehicle driving, obtain time-continuous multi-frame point cloud data collected by a laser radar.

[0129] In this step, the vehicle needs to obtain time-continuous multi-frame point cloud data collected by the lidar while driving.

[0130] In one possible implementation, to improve calibration accuracy, it's necessary to ensure the quality of the point cloud data used for extrinsic calibration. Therefore, the multi-frame point cloud data acquired and used in the calculation refers to the point cloud data collected while the vehicle is in motion and falls within a preset valid range. In practical applications, this valid range can be pre-determined based on the vehicle's height and the performance of the lidar (for example, the lidar's effective measurement range).

[0131] It is worth noting that, in this solution, the road surface on which the vehicle travels can be an ordinary road surface, and the road surface does not need to be absolutely level.

[0132] In addition, the method provided in this solution is used for offline calibration of the external parameters of the laser radar, and can also be used for online calibration of the external parameters of the laser radar, where the external parameters of the laser radar include the yaw angle, roll angle, pitch angle of the laser radar relative to the vehicle and the height of the laser radar above the ground.

[0133] For example, if the lidar external parameter machine is calibrated offline, the vehicle can be driven for a period of time, and multiple frames of point cloud data collected by the lidar during the vehicle's driving process can be obtained. Then, the multiple frames of point cloud data can be processed using an offline processing device to achieve offline calibration of the external parameters. The offline processing device can be a general-purpose computing device, including a computer device, etc.

[0134] If the external parameters of the laser radar are calibrated online, that is, the external parameters are calibrated in real time, then during the driving process of the vehicle, multi-frame point cloud data collected by the laser radar during this period of time can be obtained at intervals, and the multi-frame point cloud data corresponding to this period of time can be processed using an online processing device to realize the online calibration of the external parameters, wherein the online processing device can be a vehicle-mounted computing platform, etc.

[0135] S102: Determine the target yaw angle of the laser radar relative to the vehicle based on the multi-frame point cloud data.

[0136] In this step, the target yaw angle is calculated based on the acquired multi-frame point cloud data. Specifically, the yaw angle of the LiDAR relative to the vehicle corresponding to each frame of point cloud data can be calculated. The target yaw angle is then determined based on the yaw angle of the LiDAR relative to the vehicle corresponding to each frame of point cloud data.

[0137] S103. According to the multiple frames of point cloud data, obtain other external parameters of the laser radar corresponding to each frame of point cloud data.

[0138] In this step, a conventional algorithm can be used to process each frame of point cloud data to obtain other external parameters of the laser radar corresponding to each frame of point cloud data.

[0139] As an example, step 103 may be implemented using the following steps 3.1 to 3.2, including:

[0140] Step 3.1: For each frame of point cloud data, extract the ground normal vector corresponding to the frame of point cloud data.

[0141] Step 3.2: Calculate other external parameters of the lidar based on the ground normal vector and ideal normal vector corresponding to each frame of point cloud data.

[0142] The above steps 2.3 to 2.4 are used to process each frame of point cloud data to directly obtain other external parameters of the laser radar corresponding to each frame of point cloud data.

[0143] Specifically, in step 2.3, for each frame of point cloud data, a ground point cloud is segmented from the frame of point cloud data, and a ground plane is obtained by fitting the ground point cloud, and a ground normal vector is extracted based on the ground plane.

[0144] Optionally, a conventional algorithm in the art may be used to extract ground normal vectors from the ground point cloud, such as a Random Sample Consensus (RANSAC) algorithm, a normal vector extraction algorithm based on the least squares method, and the like.

[0145] In practical applications, in order to accurately separate the real ground point cloud and eliminate the interference of non-ground objects, when fitting the ground plane based on the ground point cloud, the interference information of non-ground objects such as surrounding curbs and shrubs in the point cloud data will be removed.

[0146] In step 2.4, for each frame of point cloud data, based on the ground normal vector and ideal normal vector corresponding to the frame of point cloud data, the transformation matrix between the ground normal vector and the ideal normal vector is calculated to solve the roll angle, pitch angle and ground clearance height of the lidar relative to the vehicle corresponding to the frame of point cloud data.

[0147] In practical applications, the normal vector obtained with the vehicle chassis as the plane can be used as the ideal normal vector of the vehicle.

[0148] Among them, other external parameters refer to other external parameters that need to be calibrated for the vehicle-mounted lidar except the yaw angle of the lidar relative to the vehicle, such as the roll angle of the lidar relative to the vehicle, the height of the lidar from the ground, etc.

[0149] In one possible implementation, the other external parameters include at least one of the roll angle of the laser radar relative to the vehicle, the pitch angle of the laser radar relative to the vehicle, and the height of the laser radar above the ground.

[0150] Specifically, the roll angle of the laser radar relative to the vehicle refers to the rotation angle of the laser radar around the x-axis of the vehicle coordinate system when collecting the frame of point cloud data; the pitch angle of the laser radar relative to the vehicle refers to the rotation angle of the laser radar around the y-axis of the vehicle coordinate system when collecting the frame of point cloud data; the ground clearance of the laser radar refers to the height of the laser radar from the ground when collecting the frame of point cloud data.

[0151] It should be noted that in actual applications, only any one of the other external parameters including the roll angle, pitch angle of the laser radar relative to the vehicle and the height of the laser radar above the ground can be calibrated, or any two of the above other external parameters can be calibrated, or the above three other external parameters can be calibrated at the same time. This application does not impose any specific restrictions on the number of other external parameters to be calibrated.

[0152] S104. Perform data statistics of other external parameters of the laser radar corresponding to each frame of point cloud data to obtain at least one intermediate statistical data.

[0153] In this step, since other external parameters of the lidar corresponding to single-frame point cloud data may be affected by noise, environmental interference, etc., there may be accidental errors. Therefore, the purpose of data statistics here is to extract intermediate statistical data that can reflect the overall trend of other external parameters from other external parameters of the lidar corresponding to multiple single-frame point cloud data through statistical methods (such as mean, median, mode, etc.).

[0154] Exemplarily, the intermediate statistical data may be a mean value determined based on other external parameters of the lidar corresponding to multiple frames of point cloud data, or the intermediate statistical data may be directly the initial data of other external parameters without computational processing.

[0155] Furthermore, in a possible implementation, step S104 may be implemented using the following steps:

[0156] The data of other external parameters of the laser radar corresponding to the multi-frame point cloud data are segmented to obtain at least one data segment corresponding to the other external parameters; the mean of each data segment is calculated, and the mean of each data segment is used as intermediate statistical data.

[0157] Among them, this application does not limit the specific method of segmentation. For example, segmentation can be performed according to the fluctuation of other external parameters of the lidar corresponding to the multi-frame point cloud data, or segmentation can be performed according to a preset number of frames, or segmentation can be performed according to a preset time length, etc.

[0158] Optionally, segmentation can be performed using the following method: The data of other extrinsic parameters of the lidar corresponding to the multi-frame point cloud data is divided into at least one data segment according to the preset variation range of the other extrinsic parameters, with the data outside the preset variation range as the segmentation point, and the data outside the preset variation range in each data segment is deleted. It should be understood that the purpose of this setting is to eliminate data with large fluctuations in the data of other extrinsic parameters, thereby eliminating the impact of road surface unevenness on the accuracy of the obtained target extrinsic parameters, and further improving the accuracy of the extrinsic parameter calibration results.

[0159] It should be understood that in actual applications, the preset range of variation is different for different external parameters. The purpose of setting the preset range of variation is to eliminate data with large fluctuations in order to obtain multiple data segments with stable values. This application does not impose specific restrictions on the setting of the preset range of variation.

[0160] In this implementation, by first dividing the data into segments and then calculating the mean of each segment, and using the mean as a means of intermediate statistical data, while retaining the local characteristics of the overall data, the single-frame error is reduced through segmented smoothing, thereby improving the reliability and accuracy of other external parameters of the target determined based on the intermediate statistical data.

[0161] S105: Perform discrete evaluation based on at least one intermediate statistical data to determine other external parameters of the laser radar target.

[0162] In this step, the intermediate statistical data must be discretely evaluated, and the target extrinsic parameters of the lidar are determined based on the evaluation results. The discrete evaluation is used to determine whether the intermediate statistical data is usable based on its degree of discreteness. Specifically, if the evaluation results indicate high quality, the target extrinsic parameters can be directly calculated based on these intermediate statistical data. If the evaluation results indicate poor quality, other methods are used to improve the reliability of the intermediate statistical data, and the target extrinsic parameters are determined based on the adjusted intermediate statistical data. For example, new extrinsic parameter data (i.e., the lidar's other extrinsic parameters) can be continuously acquired.

[0163] For example, the discrete evaluation may be performing variance analysis or aggregation analysis on a plurality of intermediate statistical data.

[0164] It should be understood that the method provided in this application is used to calibrate the lidar's yaw, roll, and pitch angles relative to the vehicle, as well as the lidar's ground clearance. The lidar's x-axis and y-axis offsets relative to the vehicle, which are also included in the lidar's extrinsic parameters, can be calibrated using common methods in the art and are not specifically limited in this solution. For example, the x-axis and y-axis offsets of the lidar relative to the vehicle can be directly measured.

[0165] It should be understood that the purpose of the analysis and processing in the above steps is to obtain the final value used for calibration for each external parameter to be calibrated based on the multiple data corresponding to the external parameter, so as to eliminate the influence of the instability of a single data on the accuracy of the external parameter calibration, thereby achieving accurate calibration of the lidar external parameters based on the point cloud data obtained when the vehicle is driving on ordinary roads.

[0166] The present embodiment provides a method for calibrating the extrinsic parameters of a vehicle-mounted lidar, which calculates the target yaw angle using multi-frame point cloud data. Simultaneously, by performing data statistics and discrete evaluation on other extrinsic parameters of the lidar corresponding to the multi-frame point cloud data, the influence of data fluctuations on ordinary roads on the calibration accuracy is eliminated, thereby ensuring the accuracy of the calibrated extrinsic parameters and achieving the effect of accurately calibrating the extrinsic parameters of the vehicle-mounted lidar on ordinary roads.

[0167] Furthermore, the second embodiment of the present application provides a method for calibrating the external parameters of a laser radar installed on a vehicle. Based on the above embodiment, this embodiment provides a specific implementation of the above step S105, including:

[0168] Step 4.1: Perform discrete evaluation based on at least one intermediate statistical data to obtain an evaluation result.

[0169] The evaluation results represent the quality of multiple current statistical data, such as stability.

[0170] Optionally, the variance of at least one intermediate statistic may be calculated and used as the evaluation result.

[0171] Step 4.2: If the evaluation result meets the preset conditions, other external parameters of the laser radar target are determined based on at least one intermediate statistical data.

[0172] In this step, it is necessary to compare whether the evaluation results meet the preset conditions, and other external parameters of the laser radar target can be directly determined based on at least one intermediate statistical data currently obtained.

[0173] Optionally, if the evaluation result is the variance of the at least one intermediate statistical data, the preset condition may be that the variance of the at least one intermediate statistic is less than the preset variance corresponding to the other external parameter. It should be understood that in actual applications, the same preset variance or different preset variances may be set for different other external parameters, wherein the smaller the preset variance, the higher the accuracy of the target other external parameter. This application does not impose any specific restrictions on the value of the preset variance corresponding to each other external parameter. For example, the preset variance may be 0.01, 0.02, 0.03, 0.05, etc.

[0174] Optionally, the following method can be used to implement “determining other target external parameters of the laser radar based on at least one intermediate statistical data”. The specific implementation method is as follows: the mean of at least one intermediate statistical data is used as other target external parameters of the laser radar.

[0175] Step 4.3: If the evaluation result does not meet the preset conditions, obtain multiple frames of new point cloud data that are time-continuous with the multiple frames of point cloud data to obtain at least one new intermediate statistical data, and determine other external parameters of the lidar target based on the at least one new intermediate statistical data and the at least one intermediate statistical data.

[0176] In this step, in order to ensure the calibration accuracy of other target external parameters, when the evaluation results do not meet the preset conditions, it can be considered that accurate other target external parameters cannot be calibrated based only on the multiple intermediate statistical numbers currently obtained. Therefore, new point cloud data can be continuously acquired to obtain at least one new intermediate statistical data corresponding to the new point cloud data, and then the target other external parameters of the lidar can be determined based on the new intermediate statistical data and at least one previous intermediate statistical data.

[0177] Optionally, if the evaluation result does not meet the preset conditions, the following steps may be used to implement "obtaining multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new intermediate statistical data, and determining other target extrinsic parameters of the lidar based on the at least one new intermediate statistical data and the at least one intermediate statistical data." The specific contents include:

[0178] Step 4.3.1: Acquire multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new data segment.

[0179] Optionally, at least one new data segment can be acquired based on the fluctuations in the data of other external parameters of the LiDAR corresponding to the new point cloud data. Exemplarily, when acquiring a new data segment, multiple frames of new point cloud data are continuously acquired while simultaneously acquiring other external parameters of the LiDAR relative to the vehicle corresponding to each frame of point cloud data. These other external parameters are then divided into new data segments. When the data of the other external parameters exceeds a preset variation range, acquisition of the data in the new data segment is stopped, and the values ​​in the new data segment that exceed the preset variation range are deleted, resulting in a final roll angle data segment. Accordingly, to acquire multiple new data segments, simply repeat the aforementioned steps.

[0180] Step 4.3.2: Calculate the mean of each new data segment and use the mean of each new data segment as the new intermediate statistical data.

[0181] Step 4.3.3: Perform discrete evaluation based on at least one new intermediate statistical data and at least one intermediate statistical data to obtain a new evaluation result.

[0182] Step 4.3.4: If the new evaluation result does not meet the preset conditions, repeat the steps until the new evaluation result meets the preset conditions, and determine other external parameters of the lidar target based on at least one intermediate statistical data and at least one new intermediate statistical data.

[0183] Optionally, when the new evaluation result meets the preset conditions, the following method can be used to implement "determining other target external parameters of the lidar based on at least one intermediate statistical data and at least one new intermediate statistical data", specifically including: taking the average of at least one intermediate statistical data and at least one new intermediate statistical data as other target external parameters of the lidar.

[0184] This implementation method continuously acquires new point cloud data in a cyclic manner until the intermediate statistical data meets the preset conditions, so as to ensure the reliability of the intermediate statistical data used to determine other external parameters of the target. At the same time, it also eliminates the impact of uneven road surface on the accuracy of external parameter calibration, thereby ensuring the accuracy of other external parameters of the target that are finally determined.

[0185] As a specific example, taking the roll angle of the laser radar relative to the vehicle as an example, the method of "calibrating the target other external parameters according to the other external parameters of the laser radar corresponding to the multi-frame point cloud data" is illustrated, which may include the following steps a to e:

[0186] Step a: divide the roll angle of the laser radar relative to the vehicle corresponding to the multi-frame point cloud data into multiple roll angle data segments according to the preset roll angle variation range, with the data where the roll angle data exceeds the preset variation range corresponding to the roll angle as the dividing point, and delete the roll angle data that exceeds the variation range corresponding to the roll angle in each roll angle data segment.

[0187] In practical applications, after obtaining a preset number of roll angle data segments, the mean variance analysis in the subsequent step be can be started to reduce the calculation amount of external parameter calibration.

[0188] Step b: for each roll angle data segment, averaging the roll angles in the roll angle data segment to obtain a mean value.

[0189] Step c: Calculate the variance of the roll angle of the laser radar relative to the vehicle based on the mean value corresponding to each roll angle data segment.

[0190] Step d: If the variance is less than the preset variance corresponding to the roll angle, then averaging the means corresponding to each roll angle data segment to obtain the target roll angle of the lidar relative to the vehicle;

[0191] Step e. If the variance is greater than or equal to the preset variance corresponding to the roll angle, new point cloud data is collected to obtain a new roll angle data segment, and the new roll angle data segment is added to multiple roll angle data segments. The above steps b to e are repeated until the target roll angle of the laser radar relative to the vehicle is obtained; wherein, the new roll angle data segment is obtained by dividing the data of the roll angle that exceeds the variation range corresponding to the roll angle as the dividing point, and deleting the data that exceeds the variation range corresponding to the roll angle.

[0192] In the above steps bd, a mean-variance analysis is performed on the multiple roll angle data segments currently obtained: the roll angle in each roll angle data segment is averaged to obtain multiple means; the variance of the multiple means currently obtained is calculated; if the obtained variance is less than the preset variance corresponding to the roll angle, it is considered that the current data amount is sufficient and the accurate target roll angle can be calibrated. At this time, the target roll angle can be obtained by averaging the multiple means currently obtained. If the obtained variance is greater than or equal to the preset variance corresponding to the roll angle, it is considered that the accurate target roll angle cannot be calibrated based on the multiple means currently obtained, and new point cloud data continues to be obtained. The new point cloud data is calculated to obtain a new roll angle data segment, and the new roll angle data segment is added to the multiple roll angle data segments previously obtained, and the mean-variance analysis is performed again until the obtained variance is less than the preset variance corresponding to the roll angle, thereby obtaining an accurately calibrated target roll angle.

[0193] It should be understood that the calculation methods for other "other external parameters" (such as the pitch angle of the laser radar relative to the vehicle, the height of the laser radar from the ground) other than the roll angle of the laser radar relative to the vehicle can be the same and will not be repeated here.

[0194] The present embodiment provides a method for calibrating the external parameters of a laser radar installed on a vehicle, which obtains an evaluation result by performing a discrete evaluation based on at least one intermediate statistical data; if the evaluation result meets the preset conditions, other target external parameters of the laser radar are determined based on the at least one intermediate statistical data; if the evaluation result does not meet the preset conditions, multiple frames of new point cloud data that are time-continuous with the multiple frames of point cloud data are obtained to obtain at least one new intermediate statistical data, and based on the at least one new intermediate statistical data and the at least one intermediate statistical data, other target external parameters of the laser radar are determined. The other target external parameters are determined based on the statistical data and the evaluation result, so as to eliminate the influence of uneven road surface on the accuracy of external parameter calibration, and realize that the laser radar external parameters can be accurately calibrated while the vehicle is driving on an ordinary road.

[0195] Furthermore, the third embodiment of the present application provides a method for calibrating the external parameters of a laser radar installed on a vehicle. Based on the above embodiment, this embodiment provides a specific implementation of the above step S102, including:

[0196] Step 2.1: Determine the coordinate system of the first frame of point cloud data in the multi-frame point cloud data as the odometry coordinate system of the lidar.

[0197] Step 2.2: Obtain the radar yaw angle of the lidar in the odometer coordinate system corresponding to each frame of point cloud data in the multi-frame point cloud data, and the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data.

[0198] In steps 2.1 and 2.2, based on the multi-frame point cloud data and the means of constructing the lidar odometry, the radar yaw angle of the lidar corresponding to each frame of point cloud data in the odometry coordinate system and the vehicle yaw angle of the vehicle in the odometry coordinate system when collecting each frame of point cloud data are obtained.

[0199] The radar yaw angle of the lidar in the odometer coordinate system refers to the rotation angle of the lidar around the Z axis of the odometer coordinate system when collecting the point cloud data of the frame.

[0200] In one possible implementation, step 2.2 may be implemented as follows:

[0201] Step 2.2.1. In the mileage coordinate system, based on multi-frame point cloud data, obtain the position change of each frame of the lidar relative to the previous frame; the position change includes the yaw angle change and translation change of the lidar.

[0202] In practical applications, a point cloud registration algorithm is used to process multi-frame point cloud data to obtain a lidar odometry. The lidar odometry includes the position change (including rotation change and translation change) of each lidar frame relative to the previous frame. The rotation change includes the change in yaw angle, roll angle and pitch angle, and the translation change includes the change in x-axis and y-axis.

[0203] Optionally, an incremental Normal Distributions Transform (NDT) method can be used based on multi-frame point cloud data to construct a lidar odometry to obtain the change in yaw angle and translation of each lidar frame relative to the previous frame. It should be understood that when using the incremental NDT algorithm to obtain lidar odometry, there is no need to extract features such as line bundles of point cloud data, so that the method in this application can be applied to both on-board mechanical lidars and solid-state lidars. Moreover, this registration algorithm, which obtains the calculation results of the current frame based on the calculation results of the previous frame, has a fast solution speed and low hardware performance requirements, making the method in this application suitable for real-time online calculation of lidar external parameters.

[0204] Step 2.2.2: Determine the radar yaw angle of the lidar in the odometry coordinate system corresponding to each frame of point cloud data based on the yaw angle change of each lidar frame relative to the previous frame.

[0205] In this step, the odometer coordinate system is used as a reference, and the yaw angle change of the lidar corresponding to each frame of cloud data in the odometer coordinate system is obtained by accumulating the yaw angle change frame by frame.

[0206] Step 2.2.3: Determine the vehicle yaw angle in the odometer coordinate system when collecting each frame of point cloud data based on the translation change of each lidar frame relative to the previous frame.

[0207] In one possible implementation, the following method may be used:

[0208] Step 2.2.3.1. Divide the multi-frame point cloud data into multiple point cloud data segments according to the preset number of frames.

[0209] In this step, in order to ensure the accuracy of the curve obtained by subsequent fitting, the acquired multi-frame point cloud data is divided into multiple point cloud data segments according to a preset number of frames. Each point cloud data segment includes a preset number of frames of time-continuous point cloud data.

[0210] In practical applications, the preset number of frames is determined in advance through testing experiments. For example, the preset number of frames may be 200 frames.

[0211] Step 2.2.3.2: In each point cloud data segment, obtain the coordinate data of the lidar in the odometry coordinate system corresponding to each frame based on the translation change of each lidar frame relative to the previous frame.

[0212] In this step, the odometry coordinate system is used as a reference, and the coordinate data (x, y) of each frame of the lidar in the target coordinate system is obtained by accumulating the translation change of the lidar frame by frame.

[0213] Step 2.2.3.3: Perform curve fitting on the coordinate data of the laser radar corresponding to each frame in the point cloud data segment to obtain a curve equation.

[0214] In this step, the coordinate data of the laser radar obtained for a preset number of frames need to be curve fitted to obtain the curve equation.

[0215] Step 2.2.3.4: For the moment when any frame of point cloud data is collected in the point cloud data segment, calculate the derivative of the curve equation with respect to the moment, and obtain the vehicle yaw angle of the vehicle in the odometer coordinate system when the frame of point cloud data is collected.

[0216] It should be understood that by solving the time derivative of the curve equation at any moment, the vehicle's yaw angle corresponding to each moment can be obtained. In this solution, by calculating the derivative of the curve equation with respect to the moment when any frame of point cloud data was collected, the vehicle's yaw angle relative to the odometer coordinate system at the time the frame of point cloud data was collected can be obtained.

[0217] This implementation method quickly obtains the vehicle's yaw angle in the odometer coordinate system by performing curve fitting on the odometer trajectory (i.e., the coordinate data of each frame of the lidar in the odometer coordinate system) and calculating the derivative of the fitted curve equation at the moment when any frame of point cloud data is collected.

[0218] Step 2.3: Based on the vehicle yaw angle and the radar yaw angle, determine the target yaw angle of the lidar relative to the vehicle.

[0219] In this step, the final target yaw angle is determined based on the vehicle yaw angle and radar yaw angle corresponding to each frame of point cloud data in the multi-frame point cloud data.

[0220] In one possible implementation, step 2.3 can be implemented using the following steps:

[0221] Step 2.3.1. For each frame of point cloud data, calculate the difference between the vehicle yaw angle and the radar yaw angle to obtain the yaw angle of the lidar relative to the vehicle.

[0222] In this step, for each frame of point cloud data, the yaw angle of the lidar relative to the vehicle corresponding to the frame of point cloud data is obtained by calculating the difference between the radar yaw angle corresponding to the frame of point cloud data and the vehicle yaw angle.

[0223] Step 2.3.2: Obtain the target yaw angle of the lidar relative to the vehicle based on the yaw angle of the lidar relative to the vehicle corresponding to each frame of point cloud data.

[0224] In this step, the target yaw angle of the laser radar relative to the vehicle is obtained by analyzing and processing the yaw angle of the laser radar relative to the vehicle corresponding to all frame point cloud data.

[0225] Optionally, step 2.3.2 can be implemented as follows:

[0226] Step 2.3.2.1. Based on the yaw angle of the vehicle relative to the lidar corresponding to each frame of point cloud data in the multiple frames of point cloud data, perform data statistics on the yaw angle of the lidar relative to the vehicle to obtain at least one intermediate statistical data;

[0227] Step 2.3.2.2 performs a discrete evaluation based on at least one intermediate statistic to determine the target yaw angle of the lidar relative to the vehicle.

[0228] It should be noted that the data statistics of the yaw angle of the laser radar relative to the vehicle, obtaining at least one intermediate statistical data and performing discrete evaluation based on at least one intermediate statistical data to determine the target yaw angle of the laser radar relative to the vehicle are the same as the data statistics and discrete estimation of other external parameters in the above embodiments, and will not be repeated here.

[0229] As a specific example, the method of “calibrating the target yaw angle based on the yaw angle of the laser radar relative to the vehicle corresponding to multiple frames of point cloud data” is described, which may include the following steps f to j:

[0230] Step f: Divide the yaw angle of the laser radar relative to the vehicle corresponding to the multi-frame point cloud data into multiple yaw angle data segments according to a preset number of frames.

[0231] In practical applications, after obtaining a preset number of yaw angle data segments, the mean-variance analysis in the subsequent step gj can be initiated to reduce the computational complexity of the external parameter calibration. The number of the preset number can be determined experimentally and is not specifically limited in this application. For example, the preset number can be 5, 10, 15, etc.

[0232] In this step, the yaw angle of the laser radar relative to the vehicle corresponding to the multi-frame point cloud data that is continuous in time is divided into multiple yaw angle data segments according to a preset number of frames. Each yaw angle data segment includes the yaw angle of the laser radar relative to the vehicle corresponding to the preset number of frames of point cloud data that are continuous in time.

[0233] It should be understood that the preset number of frames used to divide the yaw angle data segments here can be the same as the preset number of frames used to divide the point cloud data segments when obtaining the yaw angle of the vehicle based on the curve fitting method above, for example, 200 frames.

[0234] Step g: for each yaw angle data segment, average the yaw angles in the yaw angle data segment to obtain a mean value.

[0235] Step h: Calculate the variance of the yaw angle of the laser radar relative to the vehicle based on the mean value corresponding to each yaw angle data segment.

[0236] In step i, if the variance is less than the preset variance corresponding to the yaw angle, the mean corresponding to each yaw angle data segment is averaged to obtain the target yaw angle of the lidar relative to the vehicle.

[0237] In step j, if the variance is greater than or equal to the preset variance corresponding to the yaw angle, obtain a preset number of frames of new point cloud data that are continuous in time, and obtain the yaw angle of the lidar relative to the vehicle corresponding to each frame of new point cloud data as a new yaw angle data segment, add the new yaw angle data segment to multiple yaw angle data segments, and repeat the above steps b to e until the target yaw angle of the lidar relative to the vehicle is obtained.

[0238] In the above steps bd, a mean-variance analysis is performed on the multiple yaw angle data segments currently obtained: a preset number of yaw angles in each yaw angle data segment are averaged to obtain multiple means; the variance of the multiple means currently obtained is calculated; if the obtained variance is less than the preset variance corresponding to the yaw angle, it is considered that the current data volume is sufficient and the accurate target yaw angle can be calibrated. At this time, the target yaw angle can be obtained by averaging the multiple first means currently obtained. If the obtained variance is greater than or equal to the preset variance corresponding to the yaw angle, the accurate target yaw angle cannot be calibrated based on the multiple first means currently obtained. At this time, it is necessary to continue to obtain new point cloud data, calculate the new point cloud data to obtain a new yaw angle data segment, and add the new yaw angle data segment to the multiple yaw angle data segments previously obtained, and re-perform the mean-variance analysis until the obtained variance is less than the first preset variance, thereby obtaining an accurately calibrated target yaw angle.

[0239] Among them, the smaller the preset variance corresponding to the yaw angle, the higher the accuracy of the target yaw angle. This application does not impose any specific restrictions on the value of the preset variance corresponding to the yaw angle. For example, the preset variance corresponding to the yaw angle can be 0.01, 0.02, 0.03, 0.05, etc.

[0240] This implementation method eliminates the impact of the instability of individual data on the accuracy of yaw angle calibration by performing data statistics and discrete estimation on the yaw angle of the lidar relative to the vehicle corresponding to multiple frames of point cloud data, thereby achieving accurate calibration of the lidar yaw angle based on the point cloud data obtained when the vehicle is driving on ordinary roads.

[0241] In a possible implementation, before the averaging step in step c, outliers in each yaw angle data segment may be removed to further accelerate the convergence speed of the variance.

[0242] The present embodiment provides a method for calibrating the external parameters of an on-vehicle lidar, which determines the coordinate system of the first frame of point cloud data in multiple frames of point cloud data as the odometer coordinate system of the lidar; obtains the radar yaw angle of the lidar in the odometer coordinate system corresponding to each frame of point cloud data in the multiple frames of point cloud data, and the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data; and finally determines the target yaw angle of the lidar relative to the vehicle based on the vehicle yaw angle and the radar yaw angle. With the lidar odometer as the intermediary, the yaw angle of the lidar relative to the vehicle at any time is directly obtained through the radar yaw angle and the vehicle yaw angle, so that the vehicle can drive normally on ordinary roads (can turn, no need to go straight), and accurate calibration of the target yaw angle can be achieved.

[0243] Figure 2 This is a schematic diagram of the structure of the calibration device for the external parameters of the laser radar installed on the vehicle provided by this application, as shown in FIG. Figure 2 As shown, the present embodiment provides a device 20 for calibrating the external parameters of a laser radar installed on a vehicle, comprising:

[0244] An acquisition unit 201 is configured to acquire time-continuous multi-frame point cloud data collected by the laser radar during the vehicle's driving process;

[0245] A first processing unit 202 is configured to determine a target yaw angle of the laser radar relative to the vehicle based on the multiple frames of point cloud data;

[0246] The second processing unit 203 is configured to obtain other external parameters of the laser radar corresponding to each frame of point cloud data based on the multiple frames of point cloud data;

[0247] The third processing unit 204 is configured to perform data statistics of other extrinsic parameters of the laser radar according to each frame of point cloud data, to obtain at least one intermediate statistical data;

[0248] The fourth processing unit 205 is used to perform discrete evaluation based on the at least one intermediate statistical data to determine other external parameters of the target of the laser radar.

[0249] In a possible implementation, the third processing unit 204 includes:

[0250] A segmentation module, configured to segment data of other external parameters of the laser radar corresponding to the multi-frame point cloud data to obtain at least one data segment corresponding to the other external parameters;

[0251] The calculation module is used to calculate the mean of each data segment and use the mean of each data segment as the intermediate statistical data.

[0252] In a possible implementation, the fourth processing unit 205 includes:

[0253] an evaluation module, configured to perform discrete evaluation based on the at least one intermediate statistical data and obtain an evaluation result;

[0254] A first determination module is configured to determine other target extrinsic parameters of the laser radar based on the at least one intermediate statistical data if the evaluation result meets a preset condition;

[0255] The second determination module is used to obtain multiple frames of new point cloud data that are time-continuous with the multiple frames of point cloud data if the evaluation result does not meet the preset conditions, so as to obtain at least one new intermediate statistical data, and determine other external parameters of the target of the lidar based on the at least one new intermediate statistical data and the at least one intermediate statistical data.

[0256] In a possible implementation, the evaluation module is specifically configured to:

[0257] calculating a variance of the at least one intermediate statistical data, and using the variance as the evaluation result;

[0258] The preset condition is that the variance of the at least one intermediate statistic is smaller than the preset variance corresponding to the other external parameters.

[0259] In a possible implementation manner, the second determining module is specifically configured to:

[0260] Acquire multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new data segment;

[0261] Calculating a mean value of each new data segment, and using the mean value of each new data segment as the new intermediate statistical data;

[0262] performing a discrete evaluation based on the at least one new intermediate statistical data and the at least one intermediate statistical data to obtain a new evaluation result;

[0263] If the new evaluation result does not meet the preset conditions, repeat the above steps until the new evaluation result meets the preset conditions, and determine other external parameters of the target of the laser radar based on the at least one intermediate statistical data and the at least one new intermediate statistical data.

[0264] In a possible implementation, the first determining module is partially configured to:

[0265] Using the mean of the at least one intermediate statistical data as another target extrinsic parameter of the laser radar;

[0266] Accordingly, the second determining module is partially used to:

[0267] The at least one intermediate statistical data and the mean value of the at least one new intermediate statistical data are used as other target external parameters of the laser radar.

[0268] In a possible implementation, the segmentation module is specifically configured to:

[0269] The data of other external parameters of the laser radar corresponding to the multi-frame point cloud data are divided into at least one data segment according to the preset change range corresponding to the other external parameters, with the data of the other external parameters exceeding the preset change range as the dividing point, and the data exceeding the preset change range in each data segment is deleted.

[0270] In a possible implementation, the other external parameters in the second processing unit 203 include at least one of a roll angle of the laser radar relative to the vehicle, a pitch angle of the laser radar relative to the vehicle, and a height of the laser radar above the ground.

[0271] In a possible implementation, the first processing unit 202 includes:

[0272] A third determining module is used to determine the coordinate system of the first frame of point cloud data in the multiple frames of point cloud data as the odometry coordinate system of the laser radar;

[0273] an acquisition module, configured to acquire, in the multiple frames of point cloud data, a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data, and a vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data;

[0274] A fourth determination module is used to determine a target yaw angle of the laser radar relative to the vehicle based on the vehicle yaw angle and the radar yaw angle.

[0275] In a possible implementation, the acquisition module includes:

[0276] A first submodule is configured to obtain, in the mileage coordinate system, a position change of each frame of the laser radar relative to the previous frame based on the multi-frame point cloud data; wherein the position change includes a yaw angle change and a translation change of the laser radar;

[0277] The second submodule is configured to determine the radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data based on the yaw angle change of each frame of the laser radar relative to the previous frame;

[0278] The third submodule is used to determine the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data based on the translation change of each frame of the laser radar relative to the previous frame.

[0279] In a possible implementation, the third submodule is specifically configured to:

[0280] Dividing the multi-frame point cloud data into a plurality of point cloud data segments according to a preset number of frames;

[0281] In each point cloud data segment, according to the translation change of each frame of the laser radar relative to the previous frame, the coordinate data of the laser radar in the odometry coordinate system corresponding to each frame is obtained;

[0282] Performing curve fitting on the coordinate data of the laser radar corresponding to each frame in the point cloud data segment to obtain a curve equation;

[0283] For a moment when any frame of point cloud data is collected in the point cloud data segment, a derivative of the curve equation with respect to the moment is calculated to obtain a vehicle yaw angle of the vehicle in the odometer coordinate system when the frame of point cloud data is collected.

[0284] In a possible implementation, the fourth determining module includes:

[0285] A fourth submodule is configured to calculate, for each frame of point cloud data, the difference between the vehicle yaw angle and the radar yaw angle to obtain the yaw angle of the lidar relative to the vehicle;

[0286] The fifth submodule is used to obtain the target yaw angle of the laser radar relative to the vehicle based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data.

[0287] In a possible implementation manner, the fifth submodule is specifically configured to:

[0288] performing data statistics of the yaw angle of the laser radar relative to the vehicle, based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data in the multiple frames of point cloud data, to obtain at least one intermediate statistical data;

[0289] A discrete evaluation is performed based on the at least one intermediate statistic to determine a target yaw angle of the lidar relative to the vehicle.

[0290] The vehicle-mounted laser radar external parameter calibration device 20 provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effects are similar, and are not described in detail in this embodiment.

[0291] Figure 3 This is a schematic diagram of the structure of the electronic device provided in this application. Figure 3 As shown, the electronic device 30 provided in this embodiment includes: at least one processor 301 and a memory 302. Optionally, the device 30 also includes a communication component 303. The processor 301, the memory 302 and the communication component 303 are connected via a bus 304.

[0292] During the specific implementation process, at least one processor 301 executes the computer-executable instructions stored in the memory 302, so that the at least one processor 301 performs the above method.

[0293] The specific implementation process of the processor 301 can be found in the above method embodiment. Its implementation principle and technical effects are similar and will not be repeated here in this embodiment.

[0294] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly executed by a hardware processor or by a combination of hardware and software modules within the processor.

[0295] Memory can include read-only memory and random access memory. The memory can be volatile or non-volatile, or can include both volatile and non-volatile memory. Non-volatile memory can include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can include random access memory (RAM), which is used as an external cache memory. By way of example and not limitation, many forms of RAM are available. For example, static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM) and direct RAM bus random access memory (DR RAM).

[0296] A bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be categorized as address buses, data buses, and control buses. For ease of illustration, the buses in the drawings of this application are not limited to just one bus or just one type of bus.

[0297] The present application provides a vehicle, including a vehicle body and the above-mentioned electronic device, which is used to implement the above-mentioned method.

[0298] The present application also provides a computer program product, including a computer program, which implements the above method when executed.

[0299] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed, the above method is implemented.

[0300] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as SRAM, EEPROM, EPROM, PROM, ROM, magnetic storage, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0301] An exemplary readable storage medium is coupled to a processor, such that the processor can read information from and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an ASIC. Of course, the processor and the readable storage medium can also exist as discrete components in a device.

[0302] The division of units is merely a logical functional division; actual implementations may employ alternative divisions, such as combining or integrating multiple units or components into another system, or omitting or disabling certain features. Furthermore, any direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units, either through an interface, electrical, mechanical, or other means.

[0303] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0304] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0305] If a function is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (such as a personal computer, server, or network device) to perform all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, removable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0306] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0307] Finally, it should be noted that those skilled in the art will readily identify other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The present invention is not limited to the precise structure described above and illustrated in the accompanying drawings, and various modifications and variations may be made without departing from the scope thereof. The scope of the present invention is limited solely by the appended claims.

Claims

1. A method for calibrating the external parameters of a laser radar installed on a vehicle, characterized in that: include: During the driving of the vehicle, obtaining time-continuous multi-frame point cloud data collected by the laser radar; Determining a target yaw angle of the laser radar relative to the vehicle based on the multi-frame point cloud data; According to the multiple frames of point cloud data, other external parameters of the laser radar corresponding to each frame of point cloud data are obtained; Performing data statistics of other extrinsic parameters of the laser radar corresponding to each frame of point cloud data to obtain at least one intermediate statistical data; A discrete evaluation is performed based on the at least one intermediate statistical data to determine other external parameters of the target of the laser radar.

2. The method according to claim 1, characterized in that The step of performing data statistics of other external parameters of the laser radar corresponding to each frame of point cloud data to obtain at least one intermediate statistical data includes: Segmenting data of other extrinsic parameters of the laser radar corresponding to the multi-frame point cloud data to obtain at least one data segment corresponding to the other extrinsic parameters; The mean of each data segment is calculated and used as the intermediate statistical data.

3. The method according to claim 1, characterized in that The performing of discrete evaluation based on the at least one intermediate statistical data to determine other target external parameters of the laser radar includes: Performing discrete evaluation based on the at least one intermediate statistical data to obtain an evaluation result; If the evaluation result meets the preset conditions, determining other target external parameters of the laser radar based on the at least one intermediate statistical data; If the evaluation result does not meet the preset conditions, multiple frames of new point cloud data that are time-continuous with the multiple frames of point cloud data are obtained to obtain at least one new intermediate statistical data, and based on the at least one new intermediate statistical data and the at least one intermediate statistical data, other external parameters of the target of the lidar are determined.

4. The method according to claim 3, characterized in that The performing discrete evaluation on the at least one intermediate statistical data to obtain an evaluation result includes: calculating a variance of the at least one intermediate statistical data, and using the variance as the evaluation result; The preset condition is that the variance of the at least one intermediate statistic is smaller than the preset variance corresponding to the other external parameters.

5. The method according to claim 3, characterized in that The acquiring of multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new intermediate statistical data, and determining other target extrinsic parameters of the laser radar based on the at least one new intermediate statistical data and the at least one intermediate statistical data, includes: Acquire multiple frames of new point cloud data that are temporally continuous with the multiple frames of point cloud data to obtain at least one new data segment; Calculating a mean value of each new data segment, and using the mean value of each new data segment as the new intermediate statistical data; performing a discrete evaluation based on the at least one new intermediate statistical data and the at least one intermediate statistical data to obtain a new evaluation result; If the new evaluation result does not meet the preset conditions, repeat the above steps until the new evaluation result meets the preset conditions, and determine other external parameters of the target of the laser radar based on the at least one intermediate statistical data and the at least one new intermediate statistical data.

6. The method according to claim 5, characterized in that Determining other target extrinsic parameters of the laser radar based on the at least one intermediate statistical data includes: Using the mean of the at least one intermediate statistical data as another target extrinsic parameter of the laser radar; Accordingly, determining other target extrinsic parameters of the laser radar based on the at least one intermediate statistical data and the at least one new intermediate statistical data includes: The at least one intermediate statistical data and the mean value of the at least one new intermediate statistical data are used as other target external parameters of the laser radar.

7. The method according to claim 2, characterized in that The step of segmenting the data of other external parameters of the laser radar corresponding to the multi-frame point cloud data to obtain at least one data segment corresponding to the other external parameters includes: The data of other external parameters of the laser radar corresponding to the multi-frame point cloud data are divided into at least one data segment according to the preset change range corresponding to the other external parameters, with the data of the other external parameters exceeding the preset change range as the dividing point, and the data exceeding the preset change range in each data segment is deleted.

8. The method according to any one of claims 1 to 7, characterized in that The other external parameters include at least one of the roll angle of the laser radar relative to the vehicle, the pitch angle of the laser radar relative to the vehicle, and the height of the laser radar from the ground.

9. The method according to any one of claims 1 to 7, characterized in that Determining a target yaw angle of the laser radar relative to the vehicle based on the multi-frame point cloud data includes: Determine the coordinate system of the first frame of point cloud data in the multiple frames of point cloud data as the odometry coordinate system of the laser radar; Obtaining, in the multiple frames of point cloud data, a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data, and a vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data; Based on the vehicle yaw angle and the radar yaw angle, a target yaw angle of the lidar relative to the vehicle is determined.

10. The method according to claim 9, characterized in that The acquiring of the multiple frames of point cloud data, the radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data, and the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data, includes: In the mileage coordinate system, based on the multi-frame point cloud data, obtaining a position change of each frame of the laser radar relative to the previous frame; wherein the position change includes a yaw angle change and a translation change of the laser radar; Determining a radar yaw angle of the laser radar in the odometer coordinate system corresponding to each frame of point cloud data according to a change in the yaw angle of each frame of the laser radar relative to the previous frame; The vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data is determined based on the translation change of each frame of the laser radar relative to the previous frame.

11. The method according to claim 10, characterized in that Determining the vehicle yaw angle of the vehicle in the odometer coordinate system when collecting each frame of point cloud data based on the translation change of each frame of the laser radar relative to the previous frame includes: Dividing the multi-frame point cloud data into a plurality of point cloud data segments according to a preset number of frames; In each point cloud data segment, according to the translation change of each frame of the laser radar relative to the previous frame, the coordinate data of the laser radar in the odometry coordinate system corresponding to each frame is obtained; Performing curve fitting on the coordinate data of the laser radar corresponding to each frame in the point cloud data segment to obtain a curve equation; For a moment when any frame of point cloud data is collected in the point cloud data segment, a derivative of the curve equation with respect to the moment is calculated to obtain a vehicle yaw angle of the vehicle in the odometer coordinate system when the frame of point cloud data is collected.

12. The method according to claim 9, characterized in that Determining a target yaw angle of the laser radar relative to the vehicle based on the vehicle yaw angle and the radar yaw angle includes: For each frame of point cloud data, the difference between the vehicle yaw angle and the radar yaw angle is calculated to obtain the yaw angle of the lidar relative to the vehicle; According to the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data, the target yaw angle of the laser radar relative to the vehicle is obtained.

13. The method according to claim 12, characterized in that The obtaining, based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data in the multiple frames of point cloud data, a target yaw angle of the laser radar relative to the vehicle includes: performing data statistics of the yaw angle of the laser radar relative to the vehicle, based on the yaw angle of the laser radar relative to the vehicle corresponding to each frame of point cloud data in the multiple frames of point cloud data, to obtain at least one intermediate statistical data; A discrete evaluation is performed based on the at least one intermediate statistic to determine a target yaw angle of the lidar relative to the vehicle.

14. A device for calibrating the external parameters of a laser radar installed on a vehicle, characterized in that: include: An acquisition unit, configured to acquire time-continuous multi-frame point cloud data collected by the laser radar during the driving of the vehicle; a first processing unit, configured to determine a target yaw angle of the laser radar relative to the vehicle based on the multiple frames of point cloud data; A second processing unit is configured to obtain other external parameters of the laser radar corresponding to each frame of point cloud data based on the multiple frames of point cloud data; a third processing unit, configured to perform data statistics of other extrinsic parameters of the laser radar corresponding to each frame of point cloud data, to obtain at least one intermediate statistical data; A fourth processing unit is used to perform discrete evaluation based on the at least one intermediate statistical data to determine other external parameters of the target of the laser radar.

15. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 13.

16. A vehicle, characterized in that: The device comprises a vehicle body and the electronic device according to claim 15 .

17. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 13 when executed.

18. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 13 when the computer program is executed.

Images