Radar calibration method and device, electronic equipment and storage medium
By analyzing the target point cloud collected by the target radar of the target vehicle, combining the position information of the target cloud, the target transformation matrix of the target radar is determined, which solves the problems of operational complexity, high cost and low accuracy in the existing radar calibration technology, and efficient and accurate radar calibration is achieved.
Patent Information
- Application Number
- CN202311595738.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-27
- Publication Date
- 2025-06-03
AI Technical Summary
The existing radar calibration technology has problems such as complex operation, high calibration cost, and low calibration accuracy.
By collecting the target point cloud based on the target vehicle, the first transformation matrix of the target radar relative to the target vehicle is determined, and the target position is determined based on the target point cloud. Combining the angle difference between the connection line of the target position and the head direction and the measurement distance, the second transformation matrix is determined, and finally the target transformation matrix of the target radar is obtained.
It simplifies the operation process, reduces operating costs, improves calibration efficiency, and achieves accurate calibration of target radar.
Smart Images
Figure CN120085263A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of radar calibration, and in particular, to a method, device, electronic device, and storage medium for radar calibration. Background Art
[0002] Radar calibration refers to precisely calibrating a radar to ensure that it can accurately measure the distance, angle, and position of a target object. Radar calibration is a very important part of intelligent driving technology, which directly affects the measurement accuracy and reliability of the radar.
[0003] In the prior art, the methods for radar calibration mainly include offline calibration, online calibration, and adaptive calibration, etc. However, these calibration methods still have certain defects in terms of operation complexity, calibration cost, and calibration accuracy. Summary of the Invention
[0004] In view of this, the purpose of this application is to provide a method, device, electronic device, and storage medium for radar calibration to overcome the problems in the prior art.
[0005] In a first aspect, an embodiment of this application provides a method for radar calibration, and the method includes:
[0006] Determine a first transformation matrix of the target radar relative to a reference point of the target vehicle based on target point clouds collected by the target radar of the target vehicle;
[0007] Determine at least two target positions corresponding to at least two targets based on the target point clouds corresponding to the at least two targets in the target point clouds, where the at least two targets are arranged below the target radar and parallel to the transverse direction of the rear of the target vehicle, and the transverse direction of the rear is the direction perpendicular to the head orientation on the horizontal plane;
[0008] Determine a second transformation matrix based on the angle difference between the line connecting the at least two target positions and the head orientation corresponding to the target vehicle and the measured distance between the target radar and the reference point;
[0009] Determine a target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
[0010] In some technical solutions of this application, the reflection intensity of the target is different from that of other objects in the target point cloud; the target point cloud is obtained by the following method:
[0011] Screen the target point cloud according to a preset reflection intensity requirement and height requirement to obtain an overall target point cloud;
[0012] Perform clustering processing on the overall target point cloud to obtain the target point clouds corresponding to the at least two targets respectively.
[0013] In some technical solutions of the present application, the connection line of the above-mentioned at least two target positions is determined in the following manner:
[0014] Process the target point clouds corresponding to the at least two targets respectively through the random sample consensus algorithm to obtain the at least two target positions corresponding to the at least two targets respectively;
[0015] Perform connection and / or fitting processing on the at least two target positions to obtain the connection line of the at least two target positions.
[0016] In some technical solutions of the present application, the angle difference between the connection line of the above-mentioned at least two target positions and the head orientation corresponding to the target vehicle is determined in the following manner:
[0017] Determine a reference vector according to the connection line of the at least two target positions;
[0018] Determine an ideal vector according to the head orientation corresponding to the target vehicle;
[0019] Determine the angle difference according to the reference vector and the ideal vector.
[0020] In some technical solutions of the present application, determining the first transformation matrix of the target radar relative to the reference point of the target vehicle based on the target point cloud collected by the target radar of the target vehicle includes:
[0021] Perform ground correction on the target point cloud based on the initial external parameters of the target radar to obtain the first transformation matrix.
[0022] In a second aspect, an embodiment of the present application provides a device for radar calibration, and the device includes:
[0023] A first determination module, configured to determine the first transformation matrix of the target radar relative to the reference point of the target vehicle based on the target point cloud collected by the target radar of the target vehicle;
[0024] A second determination module, configured to determine at least two target positions corresponding to the at least two targets based on the target point clouds corresponding to the at least two targets among the target point clouds, wherein the at least two targets are arranged below the target radar and parallel to the transverse direction of the rear of the target vehicle, and the transverse direction of the rear is the direction perpendicular to the head orientation on the horizontal plane;
[0025] A third determination module, configured to determine a second transformation matrix based on an angle difference between a line connecting the at least two target positions and a front orientation corresponding to the target vehicle, and a measured distance between the target radar and the reference point;
[0026] A fourth determination module, configured to determine a target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
[0027] In some technical solutions of the present application, the reflection intensity of the above target is different from that of other objects in the target point cloud; the second determination module obtains the target point cloud of the target through the following method:
[0028] According to a preset reflection intensity requirement and height requirement, the target point cloud is screened to obtain an overall target point cloud;
[0029] The overall target point cloud is subjected to clustering processing to obtain the target point clouds corresponding to the at least two targets respectively.
[0030] In some technical solutions of the present application, the third determination module determines the line connecting the at least two target positions through the following method:
[0031] Through the random sample consensus algorithm, the target point clouds corresponding to the at least two targets are processed to obtain at least two target positions corresponding to the at least two targets respectively;
[0032] The at least two target positions are connected and / or fitted to obtain the line connecting the at least two target positions.
[0033] In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method for radar calibration are implemented.
[0034] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, the steps of the above method for radar calibration are executed.
[0035] The technical solutions provided by the embodiments of the present application may include the following beneficial effects:
[0036] The method of the present application includes determining a first transformation matrix of the target radar relative to a reference point of the target vehicle based on target point clouds collected by the target radar of the target vehicle; determining at least two target positions corresponding to the at least two targets based on the target point clouds corresponding to the at least two targets in the target point clouds, where the at least two targets are arranged below the target radar and parallel to the transverse direction of the rear of the target vehicle, and the transverse direction of the rear is the direction perpendicular to the head orientation on the horizontal plane; determining a second transformation matrix based on the angle difference between the line connecting the at least two target positions and the head orientation corresponding to the target vehicle and the measured distance between the target radar and the reference point; and determining a target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
[0037] In the present application, by setting targets and analyzing the target point clouds in the target point clouds collected by the target radar, the calibration of the target radar is achieved; by setting targets, there is no need to move the target vehicle, which simplifies the operation process, does not require professional personnel and professional equipment, reduces the operation cost, and improves the calibration efficiency.
[0038] To make the above objects, features, and advantages of the present application more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] To more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can be obtained based on these drawings.
[0040] Figure 1 Shows a schematic flowchart of a method for radar calibration provided by an embodiment of the present application;
[0041] Figure 2 Shows a schematic diagram of a target position provided by an embodiment of the present application;
[0042] Figure 3 Shows a schematic diagram of a specific implementation manner provided by an embodiment of the present application;
[0043] Figure 4 Shows a schematic diagram of a device for radar calibration provided by an embodiment of the present application;
[0044] Figure 5 Shows a schematic structural diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0045] To make the objectives, technical solutions and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. It should be understood that the accompanying drawings in this application only serve the purposes of illustration and description, and are not used to limit the protection scope of this application. Additionally, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate the operations implemented according to some embodiments of this application. It should be understood that the operations in the flowchart may not be implemented in sequence, and steps without a logical context relationship may be reversed or implemented simultaneously. Moreover, those skilled in the art can add one or more other operations to the flowchart or remove one or more operations from the flowchart under the guidance of the content of this application.
[0046] In addition, the described embodiments are only some embodiments of this application, rather than all embodiments. The components of the embodiments of this application described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of this application claimed, but merely represents the selected embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of this application.
[0047] It should be noted that the term "including" will be used in the embodiments of this application to indicate the existence of the features stated thereafter, but does not exclude the addition of other features.
[0048] The prior art for radar calibration mainly includes the following types:
[0049] Offline calibration: Use a pre-prepared calibration board or calibration scene, manually measure the relative position and attitude with the Lidar (radar), and then use a calibration algorithm to calculate accurate calibration parameters. This method requires a long preparation time and professional equipment, but the calibration result is relatively accurate.
[0050] Online calibration: This method performs calibration while the robot is running. The robot will automatically perform calibration by interacting with the environment. For example, during movement, the robot can monitor feature points or edge points on the ground and update the calibration parameters through an optimization algorithm. Online calibration is more flexible than offline calibration, but the calibration process may be affected by environmental changes.
[0051] Adaptive calibration: This method uses multiple Lidars for calibration, and they are calibrated with each other to improve the overall calibration accuracy. In adaptive calibration, each Lidar can observe each other and compare their respective measurement results to determine the calibration parameters and apply them to the entire system. This method usually has high robustness and accuracy.
[0052] The existing calibration methods have the following defects:
[0053] Complexity: Multi-Lidar calibration usually requires high technical requirements and professional equipment, and the calibration process is relatively complex. For non-professionals, additional training and experience may be required to perform the calibration correctly.
[0054] Time consumption: Whether it is offline calibration or online calibration, a certain amount of time is required to complete the calibration process. Especially when using multiple Lidars in a large-scale system, the calibration process may take a long time, affecting the real-time performance of practical applications.
[0055] Influence of environmental changes: Online calibration methods usually have high requirements for the environment. For scenarios with large environmental changes, such as changes in light and weather, the calibration accuracy may be affected to a certain extent.
[0056] High cost: Multi-Lidar calibration schemes usually require additional hardware devices and professional tools for calibration, which increases the overall cost of the system.
[0057] Calibration error: Even when using advanced calibration methods, due to factors such as measurement error and algorithm error, there may still be certain errors in the calibration results. These errors may have a certain impact on the overall positioning and perception capabilities of the system.
[0058] Based on this, the embodiments of the present application provide a method, device, electronic device, and storage medium for radar calibration, which will be described below through embodiments.
[0059] Figure 1 The flowchart of a method for radar calibration provided by an embodiment of the present application is shown, where the method includes steps S101 - S104; specifically:
[0060] S101. Determine a first transformation matrix of the target radar relative to a reference point of the target vehicle based on the target point cloud collected by the target radar of the target vehicle;
[0061] S102. Determine at least two target positions corresponding to the at least two targets based on the target point clouds corresponding to the at least two targets in the target point cloud, where the at least two targets are arranged below the target radar and parallel to the lateral direction of the rear of the target vehicle, and the lateral direction of the rear is the direction perpendicular to the head orientation on the horizontal plane;
[0062] S103. Determine a second transformation matrix based on the angle difference between the line connecting the at least two target positions and the head orientation corresponding to the target vehicle and the measured distance between the target radar and the reference point;
[0063] S104. Determine the target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
[0064] In the embodiment of the present application, by setting targets and analyzing the target point clouds in the target point cloud collected by the target radar, the calibration of the target radar is realized; by setting targets, there is no need to move the target vehicle, which simplifies the operation process, does not require professional personnel and professional equipment, reduces the operation cost, and improves the calibration efficiency.
[0065] Some embodiments of the present application will be described in detail below. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.
[0066] Before introducing the radar calibration method in the embodiment of the present application, the application scenario of the embodiment of the present application will be introduced first. The radar here is the radar on the target vehicle, such as a blind spot radar, etc. The blind spot lidar is mainly responsible for covering the perception space near the vehicle body, providing more accurate perception of obstacles and space around the vehicle body for the autonomous driving vehicle. The calibration of the target radar here represents the calibration of the vehicle-mounted control point and the target radar. To achieve the above calibration, at least two targets are set at the preset positions of the target vehicle in the embodiment of the present application.
[0067] When making the target, since the target needs to be regarded as a point in this application, for the convenience of data processing, the target in the embodiments of this application is selected as a circular target (of course, if the efficiency of data processing is not considered, targets of other shapes are also acceptable). Secondly, in the embodiments of this application, it is necessary to distinguish the target from other objects in the point cloud, so the reflection intensity of the target in the embodiments of this application needs to be different from that of other objects in the radar acquisition area. Further, for the convenience of distinguishing in the point cloud, the target in the embodiments of this application is a high-reflection target, that is, its reflection intensity is higher than that of other objects in the radar acquisition area. After the target is made, it needs to be set at a preset position of the target vehicle. For the convenience of subsequent data processing, the embodiments of this application select to set the target directly below the radar and parallel to the transverse direction of the rear of the target vehicle. As Figure 2 shown. In Figure 2 , 1 is the target vehicle, 2 is the target, and 3 is the target radar. A coordinate system as shown in the figure is constructed on the vehicle, where the x direction is the head-on direction of the vehicle, and the y direction is the transverse direction of the rear of the vehicle; the transverse direction of the rear of the vehicle is the direction perpendicular to the head-on direction on the horizontal plane.
[0068] In specific implementation, at least two high-reflection circular targets with a diameter of 50 cm can be made. Based on the made circular targets and relying on the normal rectangular vehicle-mounted device (target vehicle), only relying on radar data, the external parameter relationship between the static calibration vehicle-mounted control points (reference points) and the blind spot supplement radar can be completed.
[0069] S101. Based on the target point cloud collected by the target radar of the target vehicle, determine the first transformation matrix of the target radar relative to the reference point of the target vehicle.
[0070] To calibrate the target radar, the embodiments of this application adopt the method of analyzing the target point cloud collected by the target radar. By analyzing the target point cloud, the first transformation matrix of the target radar relative to the reference point of the target vehicle is accurately obtained.
[0071] When analyzing the target point cloud, first, based on the initial external parameters between the target radar and the reference point, the target point cloud is transformed into a transformed point cloud. Then, according to the preset height requirement, the transformed point cloud is screened to obtain a candidate point cloud. Next, by performing ground correction on the candidate point cloud, the first transformation matrix of the target radar relative to the reference point of the target vehicle can be determined.
[0072] In the process of screening the transformed point cloud to obtain the candidate point cloud, the transformed point cloud is screened based on a preset first height threshold to obtain the preliminarily screened point cloud. Then, based on the first height interval, the preliminarily screened point cloud is further screened to obtain the candidate point cloud within the first height interval. The first height threshold is determined according to the initial external parameters of the target radar and the reference point, while the first height interval is determined based on the height information of the preliminarily screened point cloud.
[0073] The process of ground correction for the candidate point cloud and determining the first transformation matrix of the target radar relative to the reference point of the target vehicle includes: First, the preliminary adjusted external parameters are determined according to the difference between the actual ground normal vector and the ideal ground normal vector. Then, the candidate point cloud is iteratively adjusted based on the preliminary adjusted external parameters until the ground height of the adjusted candidate point cloud is less than a preset second height threshold. During this process, the corrected height, corrected roll angle, and corrected pitch angle of the target radar relative to the reference point are determined. Finally, these correction parameters (corrected height, corrected roll angle, and corrected pitch angle) are used as the first transformation matrix.
[0074] When determining the adjusted external parameters, the adjusted roll angle and adjusted pitch angle are calculated according to the difference between the actual ground normal vector and the ideal ground normal vector, and the candidate point cloud is transformed using these adjusted angles so that the angle between the adjusted candidate point cloud and the ground is less than a preset angle threshold. Then, the height difference between the adjusted first candidate point cloud and the ground, that is, the adjusted height, is calculated. Finally, the calculated adjusted roll angle, adjusted pitch angle, and adjusted height are used as the adjusted external parameters.
[0075] The above analysis process can be implemented with reference to the Figure 3 following method: In the embodiments of this application, it is necessary to transform the target point cloud in the target radar coordinate system into the ground coordinate system. The specific transformation method is: The coordinate system of the target point cloud is transformed according to the initial external parameters of the target radar to obtain the transformed point cloud in the ground coordinate system. The initial external parameters can be determined during the production process of the target vehicle, and they represent the relative position relationship between the radar and the ground. For example, according to the initial external parameter T1_guess of the target radar, the target point cloud is transformed into the ground coordinate system according to the formula P1_base - ground = T1_guess * P1_lidar1, where P is the abbreviation of point cloud, and T is the abbreviation of transform matrix.
[0076] After obtaining the converted point cloud, the embodiments of the present application first screen the converted point cloud to exclude the points with abnormal heights, thereby obtaining a candidate point cloud. During the screening process, determining whether the point cloud meets the preset height requirement is the basis for excluding the points with abnormal heights. If a certain point cloud meets the preset height requirement, the point cloud is retained; if a certain point cloud does not meet the preset height requirement, the point cloud is excluded. In this way, those points with abnormal heights that do not meet the requirements can be excluded, thereby improving the efficiency of the subsequent division process.
[0077] The height requirement here actually includes two sub-requirements, namely the height threshold requirement and the first height interval requirement. The specific screening process is as follows: The converted point cloud is screened according to the preset height threshold to obtain the preliminarily screened point cloud, that is, the primary selected point cloud. Then, the primary selected point cloud is further screened according to the preset first height interval to obtain the candidate point cloud located within the first height interval. The present application believes that the point cloud greater than or equal to the height threshold belongs to the point cloud with abnormal height. Therefore, it is necessary to exclude the point cloud exceeding the height threshold from the converted point cloud, so as to obtain the primary selected point cloud with a height less than the height threshold. After obtaining the primary selected point cloud, it is necessary to screen it again. The basis for this screening is the first height interval, that is, the point cloud not within the first height interval should be excluded, and the point cloud located within the first height interval is used as the candidate point cloud.
[0078] It should be noted that the height threshold in the embodiments of the present application can be determined based on the initial external parameters of the target radar. After obtaining the initial external parameters of the target radar, the initial external parameters are multiplied by the preset external parameter magnification factor to obtain the height threshold. This is done to set a reasonable height threshold according to the initial external parameters, so as to effectively exclude the point clouds with abnormal heights when screening the point cloud.
[0079] The first height interval in the embodiments of the present application can be determined based on the height information of each primary selected point cloud. After obtaining the primary selected point cloud, the embodiments of the present application will calculate the average height of each primary selected point cloud. Then, by adding and subtracting the preset height from the average height respectively, the first height interval here can be obtained. This is done to set a reasonable first height interval according to the height information of the primary selected point cloud, so as to effectively exclude the point clouds not within the first height interval when screening the point cloud again.
[0080] In specific implementation, in order to filter out the points with too high heights and reduce the calculation time, P1_base - ground is sorted by height z, and the points less than 1.5 times (preset external parameter magnification factor) of the initial external parameter T1_guess are obtained (primary selected point cloud). Then, the average value of z is calculated, and adding and subtracting 10 cm (preset height) from the average value of z is used as the candidate point cloud.
[0081] After obtaining the candidate point cloud, it is necessary to identify the ground point cloud and non-ground point cloud in the candidate point cloud. In the embodiment of the present application, the method of performing ground correction on the candidate point cloud is adopted. Specifically, first calculate the actual ground normal vector of the candidate point cloud, for example, solve the actual ground normal vector of the candidate point cloud through SVD. Then, aiming at approaching the ideal ground normal vector (for example, (0, 0, 1)), adjust the actual ground normal vector of the candidate point cloud. Here, the adjustment is based on the adjustment external parameters of the candidate point cloud.
[0082] The adjustment external parameters of the candidate point cloud include roll angle, pitch angle and ground height. Among them, roll and pitch are obtained by converting the difference between the actual ground normal vector and the ideal ground normal vector of the candidate point cloud, while the ground height is obtained by calculating the average height of the candidate point cloud. Specifically, determine the roll angle and pitch angle based on the actual ground normal vector and the ideal ground normal vector; then, adjust the candidate point cloud based on these roll angles and pitch angles so that the angle between the adjusted candidate point cloud and the ground is less than a preset angle threshold; then, calculate the height of the adjusted candidate point cloud from the ground; finally, take the roll angle, pitch angle and ground height as the adjustment external parameters.
[0083] In this way, the candidate point cloud can be ground-corrected, so as to better identify the ground point cloud and non-ground point cloud. At the same time, this method can also improve the accuracy and efficiency of point cloud division.
[0084] When adjusting the actual ground normal vector of the candidate point cloud, continuous iteration is required until the ground height of the adjusted point cloud is less than a preset target height threshold. This target height threshold is set according to the actual height of the ground, representing the ideal height range. When the ground height of the adjusted point cloud is less than this preset target height threshold, it indicates that the actual ground normal vector of the candidate point cloud has approached or coincided with the ideal ground normal vector, that is, after adjustment, the candidate point cloud is very close to the actual ground.
[0085] Next, it is also necessary to screen the adjusted candidate point cloud again, this time based on the second height interval. The second height interval is determined according to the adjustment external parameters of the target radar. Specifically, by adding or subtracting a preset height to the adjustment external parameters of the target radar respectively, the second height interval can be obtained. The points within this height interval are considered non-ground point clouds, while the points not within this height interval are considered ground point clouds.
[0086] In specific implementation, the singular value decomposition (SVD) is used to solve the ground normal vector normal1_svd and the height h1_svd of the candidate point cloud. The difference is calculated between the normal vector normal1_svd solved by SVD and the ideal ground normal vector normal_ground (0, 0, 1), and the difference is converted into the roll angle roll1 and the pitch angle pitch1. The ideal ground plane normal vector is the (0, 0, 1) vector. The vector difference is converted into a rotation matrix, and then the rotation matrix is converted into the roll angle roll and the pitch angle pitch. The adjusted variables of the roll angle roll and the pitch angle pitch are calculated, and the ground point cloud is adjusted using the variables so that the ground point cloud conforms to the real ground. The average height H1_ground - everage of the converted ground point cloud is calculated, and the lidar height above the ground H1_fix = -H1_ground - everage is obtained. The adjusted roll1, pitch1, and H1_fix are fed back as the external parameter T1_plane - fit to adjust the input point cloud. This is done until the adjusted value of H1_fix is less than 0.5 cm. (The point cloud is pre - processed using the adjusted external parameter T1_plane - fit and transformed to make the point cloud more parallel to the ground. A new optimized external parameter T1_plane - fit will be obtained in each iteration). After the ground plane correction, the height Z_bl, roll angle Roll_bl, and pitch angle Pitch_bl of the lidar relative to the base are obtained. At the same time, it is converted into the transformation matrix T_bl(plane).
[0087] S102. Based on the target point clouds corresponding to at least two targets in the target point cloud, determine at least two target positions corresponding to the at least two targets, where the at least two targets are arranged below the target radar and parallel to the transverse direction of the rear of the target vehicle, and the transverse direction of the rear is the direction perpendicular to the head - on direction on the horizontal plane.
[0088] In the embodiment of the present application, the amount of target data is at least two, because at least two points are required to determine a straight line. It can also be more than two targets. For example, three targets, four targets, five targets, etc. can be used.
[0089] In order to use the target point clouds of the target points on each target, the embodiments of the present application need to distinguish the target point cloud and determine the target point clouds of each target from the target point cloud. Since when placing the target, the embodiments of the present application place the target directly below the target radar, and the reflection intensity of the target is different from that of other objects, the screening of the target point cloud is based on the reflection intensity requirement and the height requirement. Since the reflection intensity of the target in the embodiments of the present application is relatively high, the reflection intensity requirement in the embodiments of the present application is greater than or equal to the reflection intensity threshold, that is, the point cloud with a reflection intensity greater than or equal to the reflection intensity threshold is screened out. Since the target in the embodiments of the present application is placed on the ground, its height is relatively high. The height requirement here is greater than or equal to the height threshold, that is, the point cloud with a height greater than or equal to the height threshold is screened out. The point cloud that simultaneously meets the height requirement and the reflection intensity requirement is used as the target point cloud.
[0090] Since there are multiple targets in the embodiments of the present application, the target point clouds screened out here are the point clouds corresponding to multiple targets. After obtaining the target point clouds of all targets, further distinction is required to separately distinguish the point clouds of each target. When distinguishing the overall target point cloud, the embodiments of the present application adopt a clustering method. By performing clustering processing on the overall target point cloud, the target point clouds corresponding to each target are obtained. For example, two target point clouds are clustered, namely target 1 P_target1 and target 2 P_target2.
[0091] S103. Determine a second transformation matrix based on the angle difference between the line connecting the at least two target positions and the heading direction of the target vehicle corresponding thereto, and the measurement distance between the target radar and the reference point.
[0092] After obtaining the target point clouds of each target, the embodiments of the present application need to process the target as a point, so it is necessary to determine the target positions of each target point cloud. To facilitate the determination of the target position, the embodiments of the present application select circular targets. The process of determining the target position is to process each of the target point clouds using the random sample consensus algorithm. After obtaining the positions of each target, the positions of each target are connected and / or fitted to obtain the line connecting at least two target positions. After determining the line connecting the target positions, it is also necessary to calculate the angle difference between the line and the heading direction of the target vehicle corresponding thereto. For example, the target positions of target 1 P_target1 and target 2 P_target2 are P_center1 and P_center2 respectively, that is, the angle difference between the line connecting P_center1 and P_center2 and Figure 2 the x direction.
[0093] Specifically, a reference vector is generated according to the connection line of at least the determined target positions, an ideal vector is determined according to the heading direction of the target vehicle, and the angle difference is calculated according to the reference vector and the ideal vector. For example, the generated reference vector is V_12, and V_12 is a vector generated in the xy plane according to P_center1 and P_center2. The generated ideal vector is the positive x-direction vector (0, 1), denoted as V_x. The calculated angle difference is R_angle:
[0094] R_angle = arccos[(V_12 * V_x) / (|V_12| * |V_x|)].
[0095] After obtaining the angle difference, it is also necessary to measure the distance between the target radar and the reference point. The distances here include Figure 2 the distances in the x direction, y direction, and z direction shown in the figure, denoted as X_base_lidar, Y_base_lidar, and Z_base_lidar respectively. Transforming X_base_lidar, Y_base_lidar, and Z_base_lidar with R_angle gives the second transformation matrix T_bl(target).
[0096] S104. Determine the target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
[0097] After obtaining the first transformation matrix and the second transformation matrix, the embodiments of the present application fuse the first transformation matrix and the second transformation matrix to obtain the target transformation matrix of the target radar relative to the reference point of the target vehicle. The specific fusion here is as follows:
[0098] T_bl = T_bl(plane) * T_bl(target). Where T_bl is the target transformation matrix.
[0099] Figure 4 The figure shows a schematic structural diagram of a radar calibration device provided by the embodiments of the present application. The device includes:
[0100] A first determination module, configured to determine a first transformation matrix of the target radar relative to a reference point of the target vehicle based on target point clouds collected by a target radar of the target vehicle;
[0101] A second determination module, configured to determine at least two target positions corresponding to at least two targets based on target point clouds corresponding to the at least two targets in the target point clouds, where the at least two targets are arranged below the target radar and parallel to the transverse direction of the rear of the target vehicle, and the transverse direction of the rear is the direction perpendicular to the heading direction on the horizontal plane;
[0102] A third determination module, configured to determine a second transformation matrix based on an angle difference between a line connecting the at least two target positions and a vehicle head orientation corresponding to the target vehicle, and a measured distance between the target radar and the reference point;
[0103] A fourth determination module, configured to determine a target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
[0104] The target has a different reflection intensity from other objects in the target point cloud, and the target is disposed directly below the target radar;
[0105] The apparatus obtains the target point cloud of the target through the following method:
[0106] Filter the target point cloud according to a preset reflection intensity requirement and height requirement to obtain an overall target point cloud;
[0107] Cluster the overall target point cloud to obtain the target point cloud of each target.
[0108] The apparatus determines the line connecting the at least two target positions through the following method:
[0109] Process each target point cloud through a random sample consensus algorithm to obtain the target position of each target point cloud;
[0110] Connect the target positions of each target to obtain the line connecting the at least two target positions.
[0111] The apparatus determines the angle difference through the following method:
[0112] Generate a reference vector according to the determined line connecting the at least two target positions;
[0113] Determine an ideal vector according to the vehicle head orientation corresponding to the target vehicle;
[0114] Calculate the angle difference according to the reference vector and the ideal vector.
[0115] The apparatus determines the first transformation matrix through the following method:
[0116] Perform ground correction on the target point cloud based on the initial extrinsic parameters of the target radar to obtain the first transformation matrix.
[0117] Such as Figure 5As shown, an embodiment of the present application provides an electronic device for implementing the radar calibration method in the present application. The device includes a memory, a processor, a bus, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned radar calibration method are implemented.
[0118] Specifically, the above-mentioned memory and processor can be general memory and processor, and no specific limitation is made here. When the processor runs the computer program stored in the memory, it can execute the above-mentioned radar calibration method.
[0119] Corresponding to the radar calibration method in the present application, an embodiment of the present application also provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium, and when the computer program is run by a processor, the steps of the above-mentioned radar calibration method are executed.
[0120] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, etc. When the computer program on the storage medium is run, it can execute the above-mentioned radar calibration method.
[0121] In the embodiments provided by the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. The system embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some communication interfaces. The indirect coupling or communication connection of the system or unit can be in an electrical, mechanical or other form.
[0122] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0123] In addition, each functional unit in the embodiments provided by the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.
[0124] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.
[0125] It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.
[0126] Finally, it should be noted that: the above-mentioned embodiments are only specific implementation manners of this application, used to illustrate the technical solution of this application, rather than limiting it. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: any person skilled in the art within the technical scope disclosed in this application can still modify the technical solutions recorded in the foregoing embodiments, or can easily think of changes, or make equivalent replacements for some of the technical features; and these modifications, changes, or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.
Claims
1. A method for radar calibration, characterized in that, the method includes: Based on the target point cloud collected by the target radar of the target vehicle, determining a first transformation matrix of the target radar relative to the reference point of the target vehicle; Based on the target point clouds corresponding to at least two targets in the target point cloud, determining at least two target positions corresponding to the at least two targets, wherein the at least two targets are arranged below the target radar and parallel to the transverse direction of the rear of the target vehicle, and the transverse direction of the rear is the direction perpendicular to the head orientation on the horizontal plane; Based on the angle difference between the line connecting the at least two target positions and the head orientation corresponding to the target vehicle and the measured distance between the target radar and the reference point, determining a second transformation matrix; Based on the first transformation matrix and the second transformation matrix, determining a target transformation matrix corresponding to the target radar.
2. The method according to claim 1, characterized in that, the reflection intensity of the target is different from that of other objects in the target point cloud; the target point cloud is obtained by the following method: According to the preset reflection intensity requirement and height requirement, screening the target point cloud to obtain an overall target point cloud; Performing clustering processing on the overall target point cloud to obtain the target point clouds corresponding to the at least two targets respectively.
3. The method according to claim 1, characterized in that, the line connecting the at least two target positions is determined by the following method: By using the random sample consensus algorithm, processing the target point clouds corresponding to the at least two targets respectively to obtain the at least two target positions corresponding to the at least two targets respectively; Connecting and / or fitting the at least two target positions to obtain the line connecting the at least two target positions.
4. The method according to claim 1, characterized in that, the angle difference between the line connecting the at least two target positions and the head orientation corresponding to the target vehicle is determined by the following method: According to the line connecting the at least two target positions, determining a reference vector; According to the head orientation corresponding to the target vehicle, determining an ideal vector; According to the reference vector and the ideal vector, determining the angle difference.
5. The method according to claim 1, characterized in that, Based on the target point cloud collected by the target radar of the target vehicle, determining a first transformation matrix of the target radar relative to the reference point of the target vehicle includes: Based on the initial external parameters of the target radar, performing ground correction on the target point cloud to obtain the first transformation matrix.
6. A device for radar calibration, characterized in that, the device includes: A first determination module, configured to determine a first transformation matrix of the target radar relative to the reference point of the target vehicle based on the target point cloud collected by the target radar of the target vehicle; A second determination module, configured to determine at least two target positions corresponding to the at least two targets based on the target point clouds corresponding to the at least two targets disposed below the target radar and parallel to the lateral direction of the rear of the target vehicle, where the lateral direction of the rear is the direction perpendicular to the head orientation on a horizontal plane; A third determination module, configured to determine a second transformation matrix based on the angle difference between the line connecting the at least two target positions and the head orientation corresponding to the target vehicle and the measured distance between the target radar and the reference point; A fourth determination module, configured to determine a target transformation matrix corresponding to the target radar based on the first transformation matrix and the second transformation matrix.
7. The apparatus according to claim 6, wherein, the reflection intensity of the target is different from that of other objects in the target point cloud; the second determination module obtains the target point cloud in the following manner: screen the target point cloud according to a preset reflection intensity requirement and height requirement to obtain an overall target point cloud; perform clustering processing on the overall target point cloud to obtain the target point clouds corresponding to the at least two targets respectively.
8. The apparatus according to claim 6, wherein, the third determination module determines the line connecting the at least two target positions in the following manner: process the target point clouds corresponding to the at least two targets respectively through a random sample consensus algorithm to obtain the at least two target positions corresponding to the at least two targets respectively; perform connection and / or fitting processing on the at least two target positions to obtain the line connecting the at least two target positions.
9. An electronic device, wherein, comprises: a processor, a memory and a bus, the memory stores machine-readable instructions executable by the processor, when the electronic device runs, the processor communicates with the memory through the bus, and when the machine-readable instructions are executed by the processor, the steps of the radar calibration method according to any one of claims 1 to 7 are executed.
10. A computer-readable storage medium, wherein, a computer program is stored on the computer-readable storage medium, and when the computer program is run by a processor, the steps of the radar calibration method according to any one of claims 1 to 7 are executed.