External parameter calibration method and system

Abstract

The present application discloses an external parameter calibration method and system, which belongs to the field of autonomous driving technology. The method includes: scanning a circular hole calibration plate by a solid-state radar and a mechanical radar to obtain solid-state radar point cloud data and mechanical radar point cloud data; performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud; performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to determine the center position of the measured circular hole; performing verification, and determining the data that meets the verification conditions as the registration circular hole position data; solving the external parameters by the registration circular hole position data to obtain the external parameters. The present application calibrates the solid-state radar and the mechanical radar by using a circular hole calibration plate in their common viewing area, and calibrates the external parameters by detecting the circular hole calibration plate, thereby improving the accuracy of the external parameter calibration and improving the reliability of multi-radar fusion.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to an external parameter calibration method and system. Background Art

[0002] The perception module in the autonomous driving field is responsible for acquiring and interpreting surrounding scene information. Its performance and reliability directly impact downstream control chains such as positioning, path planning, and decision-making. As one of the core sensors in the autonomous driving perception module, lidar (LiDAR) provides three-dimensional coordinate information of objects with itself as the reference origin. However, due to disadvantages such as vertical blind spots in mechanical rotating radars and the small field of view of solid-state radars, single-use radars have low information density. Therefore, it has been proposed to combine the two types of radars to achieve wider scene coverage. However, the differences in the laser scanning methods of the two types of radars have a certain impact on the accuracy of the external parameter calibration of the two radars, which is a particularly difficult problem in the field of autonomous driving technology.

[0003] The typical solution currently involves using a traditional registration algorithm to calculate the coordinate transformation matrix between the two radars based on the point clouds in the common view area of ​​the two radars. The point cloud registration accuracy of the normal distribution transformation algorithm is closely related to the size of the voxel grid, requiring a trade-off between accuracy and algorithm real-time performance. Furthermore, this method suffers from low accuracy in point cloud registration with small Field of View (FoV). The iterative closest point algorithm optimizes the coordinate transformation matrix using the distance between the registration points in the two lidar point clouds as a cost function. However, two lidar scanned point clouds rarely detect the same object at the same time, so the registration points cannot describe the same object location. The algorithm's principles indicate that registration accuracy will be limited. Furthermore, this method requires a good initial value to prevent the solution from entering a local optimum. Therefore, it is imperative to develop a high-precision extrinsic calibration method for solid-state and mechanical radars. Summary of the Invention

[0004] The present application proposes an external parameter calibration method, device and equipment to solve the problem that when a solid-state radar and a mechanical radar are used together, the alignment accuracy of the two radars is low, and the accuracy is poor when the two radars are calibrated together.

[0005] In a first aspect, a technical solution of the present application provides an external parameter calibration method, comprising: when a carrier is stationary, scanning a circular hole calibration plate in a common viewing area by a solid-state radar and a mechanical radar, respectively, to obtain solid-state radar point cloud data and mechanical radar point cloud data, respectively; performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data, respectively, to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate; performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud, respectively, to determine the measured circular hole center position; verifying the measured circular hole center position according to the actual circular hole center position in the circular hole calibration plate, and determining the data that meets the verification conditions as the registered circular hole position data; solving the external parameters through multiple sets of registered circular hole position data, and obtaining the corresponding external parameters of the solid-state radar and the mechanical radar.

[0006] Optionally, the circular hole calibration plate in the common view area is scanned by a solid-state radar and a mechanical radar respectively to obtain solid-state radar point cloud data and mechanical radar point cloud data respectively, including: scanning the circular hole calibration plate multiple times by the solid-state radar to obtain multiple frames of solid-state radar original point cloud data; performing straight-through filtering on the solid-state radar original point cloud data, extracting the area of ​​interest corresponding to the circular hole calibration plate, and obtaining solid-state radar point cloud data.

[0007] Optionally, the solid-state radar and the mechanical radar are respectively scanned by the solid-state radar and the mechanical radar to obtain the solid-state radar point cloud data and the mechanical radar point cloud data respectively, and also includes: scanning the circular hole calibration plate by the mechanical radar once to obtain the mechanical radar original point cloud data; extracting the area of ​​interest corresponding to the circular hole calibration plate from the mechanical radar original point cloud data to obtain the mechanical radar point cloud data.

[0008] Optionally, the solid-state radar point cloud data and the mechanical radar point cloud data are segmented respectively to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate, including selecting three point cloud points with a preset threshold number of times in the solid-state radar point cloud data or the mechanical radar point cloud data, and constructing corresponding multiple fitting planes respectively; determining the point cloud data in the solid-state radar point cloud data or the mechanical radar point cloud data whose distance to the fitting plane is less than the preset distance of the plane as the predetermined plane point cloud; determining the number of point cloud points in the multiple predetermined plane point clouds, and determining the predetermined plane point cloud with the largest number of point cloud points as the corresponding solid-state radar plane point cloud or the mechanical radar plane point cloud.

[0009] Optionally, circular hole edge detection is performed on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to determine the measured circular hole center position, including: performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to obtain corresponding circular hole edge point clouds; projecting the circular hole edge point cloud onto the plane where the circular hole calibration plate is located to obtain two-dimensional circular hole edge points, and then obtaining the measured circular hole center position.

[0010] Optionally, circular hole edge detection is performed on the mechanical radar plane point cloud respectively to obtain the corresponding circular hole edge point cloud, including: storing the distance information of each point cloud point in the mechanical radar plane point cloud in the form of a wire harness; judging whether the distance between the point cloud point corresponding to each wire harness and its left and right field points exceeds a preset distance threshold; if the distance is greater than the preset distance threshold and the point cloud point meets the global edge point feature, the point cloud point is determined as an edge point.

[0011] Optionally, the circular hole edge point cloud is projected onto the plane where the circular hole calibration plate is located to obtain two-dimensional circular hole edge points, and then the measured circular hole center position is obtained, including: using a two-dimensional circle model, sampling a preset number of times in the two-dimensional circular hole edge points, and sampling three edge points each time, fitting a preset number of circular hole edge lines, wherein the three edge points are not sampled repeatedly; the two-dimensional circular hole edge points whose distance to the circular hole edge line is less than a preset distance threshold are determined as the circular hole inner points of the corresponding circular hole edge line; the number of circular hole inner points of each circular hole edge line is determined, and the circular hole edge line with the most circular hole inner points is determined as the circular hole edge line of the circular hole calibration plate, and then the measured circular hole center position is determined.

[0012] Optionally, the measured circular hole center position is verified according to the actual circular hole center position in the circular hole calibration plate, and the one that meets the verification conditions is determined as the registration circular hole position data, including: calculating the distance error between multiple measured circular hole center positions and the corresponding multiple actual circular hole center positions. If the distance error is less than a preset threshold, the measured circular hole center position is determined as the registration circular hole position data.

[0013] Optionally, determining the correspondence between the measured circular hole center position and the real circular hole center position includes: determining, in the radar coordinate system, the angle between the line connecting the center position of each measured circular hole and the coordinate origin and the xy plane; comparing each angle, wherein the measured circular hole center position corresponding to the larger angle corresponds to the real circular hole center position at the upper part of the circular hole calibration plate, and the measured circular hole center position corresponding to the smaller angle corresponds to the real circular hole center position at the lower part of the circular hole calibration plate.

[0014] Optionally, determining the correspondence between the measured circular hole center position and the real circular hole center position includes: determining the projection points of the center positions of each measured circular hole on the xy plane in the radar coordinate system; comparing the positions of the projection points, wherein the measured circular hole center position corresponding to the projection point on the left corresponds to the real circular hole center position on the left side of the circular hole calibration plate, and the measured circular hole center position corresponding to the projection point on the right corresponds to the real circular hole center position on the right side of the circular hole calibration plate.

[0015] In a second aspect, in a technical solution of the present application, an external parameter calibration system is provided, comprising:

[0016] When the carrier is stationary, the solid-state radar and the mechanical radar are used to scan the circular hole calibration plate in the common view area respectively, and the solid-state radar point cloud data and the mechanical radar point cloud data are obtained respectively;

[0017] Performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data respectively, and determining modules of the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate;

[0018] Performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively, and determining a module for measuring the center position of the circular hole;

[0019] Verifying the center position of the measured circular hole according to the center position of the real circular hole in the circular hole calibration plate, and determining the module that meets the verification conditions as the registration circular hole position data;

[0020] A module is provided for calculating external parameters by using multiple sets of the registration circular hole position data to obtain corresponding external parameters of the solid-state radar and the mechanical radar.

[0021] Optionally, the device is used to implement the external parameter calibration method in the first aspect.

[0022] In a third aspect, in a technical solution of the present application, a computer-readable storage medium is provided, which stores computer instructions, wherein the computer instructions are operated to execute the external parameter calibration method in the first aspect.

[0023] In a fourth aspect, in a technical solution of the present application, a computer device is provided, which includes a processor and a memory, the memory storing computer instructions, wherein: the processor operates the computer instructions to execute the external parameter calibration method in the first aspect solution.

[0024] The beneficial effects of the present application are as follows: the present application calibrates the solid-state radar and the mechanical radar through a circular hole calibration plate in the common viewing area of ​​the solid-state radar and the mechanical radar, and calibrates the external parameters by detecting the circular hole calibration plate, thereby improving the accuracy of the external parameter calibration and improving the reliability of multi-laser fusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] To more clearly illustrate the technical solutions in the embodiments of this application, the following briefly introduces the drawings required for describing the embodiments. The drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0026] Figure 1 This is a flow chart of an implementation method of the external reference calibration method of the present application;

[0027] Figure 2 It is a flow chart of an example of the external reference calibration method of this application;

[0028] Figure 3 It is a structural diagram of an implementation scheme of the external parameter calibration system of the present application. DETAILED DESCRIPTION

[0029] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0030] The terms "first," "second," "third," "fourth," and the like (if any) in the specification and claims of the present application and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or sequential sequence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present application described herein can, for example, be implemented in orders other than those illustrated or described herein. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions, e.g., a product or apparatus comprising a series of steps or units is not necessarily limited to those units explicitly listed, but may include other units not explicitly listed or inherent to those products or apparatuses.

[0031] The perception module in the autonomous driving field is responsible for acquiring and interpreting surrounding scene information. Its performance and reliability directly impact downstream control chains such as positioning, path planning, and decision-making. As one of the core sensors in the autonomous driving perception module, lidar (LiDAR) provides three-dimensional coordinate information of objects with itself as the reference origin. However, due to disadvantages such as vertical blind spots in mechanical rotating radars and the small field of view of solid-state radars, single-use radars have low information density. Therefore, it has been proposed to combine the two types of radars to achieve wider scene coverage. However, the differences in the laser scanning methods of the two types of radars have a certain impact on the accuracy of the external parameter calibration of the two radars, which is a particularly difficult problem in the field of autonomous driving technology.

[0032] The typical solution currently involves using a traditional registration algorithm to calculate the coordinate transformation matrix between the two radars based on the point clouds in the common view area of ​​the two radars. The point cloud registration accuracy of the normal distribution transformation algorithm is closely related to the size of the voxel grid, requiring a trade-off between accuracy and algorithm real-time performance. Furthermore, this method suffers from low accuracy in point cloud registration with small Field of View (FoV). The iterative closest point algorithm optimizes the coordinate transformation matrix using the distance between the registration points in the two lidar point clouds as a cost function. However, two lidar scanned point clouds rarely detect the same object at the same time, so the registration points cannot describe the same object location. The algorithm's principles indicate that registration accuracy will be limited. Furthermore, this method requires a good initial value to prevent the solution from entering a local optimum. Therefore, it is imperative to develop a high-precision extrinsic calibration method for solid-state and mechanical radars.

[0033] In response to the above problems, the present application proposes an external parameter calibration method, comprising: when the carrier is stationary, scanning the circular hole calibration plate in the common viewing area by means of a solid-state radar and a mechanical radar, respectively, to obtain solid-state radar point cloud data and mechanical radar point cloud data, respectively; performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data, respectively, to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate; performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud, respectively, to determine the measured circular hole center position; verifying the measured circular hole center position according to the actual circular hole center position in the circular hole calibration plate, and determining the data that meets the verification conditions as the registered circular hole position data; solving the external parameters through multiple sets of registered circular hole position data, and obtaining the corresponding external parameters of the solid-state radar and the mechanical radar.

[0034] The external parameter calibration method of the present application calibrates the solid-state radar and the mechanical radar through a circular hole calibration plate in the common viewing area of ​​the solid-state radar and the mechanical radar, and calibrates the external parameters by detecting the circular hole calibration plate, thereby improving the accuracy of the external parameter calibration and the reliability of multi-laser fusion.

[0035] 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.

[0036] Figure 1 An embodiment of the external parameter calibration method of the present application is shown.

[0037] exist Figure 1In the illustrated embodiment, the external parameter calibration method of the present application includes process S101, in which, when the carrier is stationary, the circular hole calibration plate in the common viewing area is scanned by the solid-state radar and the mechanical radar respectively to obtain solid-state radar point cloud data and mechanical radar point cloud data, respectively.

[0038] In this embodiment, when performing external parameter calibration of the solid-state radar and the mechanical radar, first the carriers of the two radars need to be stationary. For example, the carrier can be a collection vehicle, etc. Then the solid-state radar and the mechanical radar respectively scan the circular hole calibration plate in the common viewing area of ​​the two to obtain the corresponding solid-state radar point cloud data and mechanical radar point cloud data.

[0039] Specifically, the number of circular holes on the circular hole calibration plate can be reasonably set according to actual needs, and at least two circular holes are included.

[0040] Optionally, the circular hole calibration plate in the common view area is scanned by a solid-state radar and a mechanical radar respectively to obtain solid-state radar point cloud data and mechanical radar point cloud data respectively, including: scanning the circular hole calibration plate multiple times by the solid-state radar to obtain multiple frames of solid-state radar original point cloud data; performing straight-through filtering on the solid-state radar original point cloud data, extracting the area of ​​interest corresponding to the circular hole calibration plate, and obtaining solid-state radar point cloud data.

[0041] In this optional embodiment, since solid-state radar and mechanical radar belong to different radar types, the point cloud data obtained by scanning the circular hole calibration plate is also acquired through corresponding scanning methods. Solid-state radar has the characteristic of non-repetitive scanning, so to obtain accurate point cloud data of the circular hole calibration plate, multiple scans are required, ultimately generating multiple frames of solid-state radar raw point cloud data. After accumulating multiple frames of point cloud data, the obtained solid-state radar raw point cloud data is subjected to straight-through filtering to remove erroneous point clouds, and the point cloud corresponding to the circular hole calibration plate's region of interest is extracted to obtain the solid-state radar point cloud data.

[0042] Optionally, the solid-state radar and the mechanical radar are respectively scanned by the solid-state radar and the mechanical radar to obtain the solid-state radar point cloud data and the mechanical radar point cloud data respectively, and also includes: scanning the circular hole calibration plate by the mechanical radar once to obtain the mechanical radar original point cloud data; extracting the area of ​​interest corresponding to the circular hole calibration plate from the mechanical radar original point cloud data to obtain the mechanical radar point cloud data.

[0043] In this optional embodiment, since the point cloud scanning position of the mechanical rotating radar is the same when it is stationary, there is no need to accumulate multiple frames of point cloud. Therefore, after a single scan of the circular hole calibration plate, the original point cloud data of the mechanical radar is obtained, and then the area of ​​interest corresponding to the circular hole calibration plate is extracted to obtain the mechanical radar point cloud data related to the external parameter calibration.

[0044] exist Figure 1In the illustrated embodiment, the external parameter calibration method of the present application includes process S102, which performs plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data, respectively, to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate.

[0045] In this implementation, after scanning and region-of-interest extraction, solid-state radar point cloud data and mechanical radar point cloud data are obtained. Planar segmentation is then performed on each of these data, ultimately determining the plane point cloud corresponding to the circular hole calibration plate within the solid-state and mechanical radar point cloud data for subsequent detection.

[0046] Optionally, the solid-state radar point cloud data and the mechanical radar point cloud data are segmented respectively to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate, including: selecting three point cloud points with a preset threshold number of times in the solid-state radar point cloud data or the mechanical radar point cloud data, and constructing corresponding multiple fitting planes respectively; determining the point cloud data in the solid-state radar point cloud data or the mechanical radar point cloud data whose distance to the fitting plane is less than the preset distance of the plane as the predetermined plane point cloud; determining the number of point cloud points in the multiple predetermined plane point clouds, and determining the predetermined plane point cloud with the largest number of point cloud points as the corresponding solid-state radar plane point cloud or the mechanical radar plane point cloud.

[0047] In this optional embodiment, the following example illustrates the determination of a solid-state radar plane point cloud from solid-state radar point cloud data. First, a preset threshold number of selections are performed on the solid-state radar point cloud data, with three point cloud points selected each time. A plane is then determined using these three point cloud points. Then, using this plane as a criterion, point clouds whose distance from the fitted plane is less than a preset distance from the plane are extracted from the solid-state radar point cloud data as the predetermined plane point clouds, ultimately obtaining the predetermined plane point clouds corresponding to the preset threshold number of points. The number of point cloud points in the multiple predetermined plane point clouds is then determined, and the predetermined plane point cloud with the largest number of point cloud points is determined as the corresponding solid-state radar plane point cloud. Specifically, during actual plane segmentation and determination of a solid-state radar plane point cloud or a mechanical radar plane point cloud, three point cloud points are randomly selected. Based on the coordinates of the point cloud points in the radar coordinate system, the plane equation ax + by + cz + d = 0 is constructed. Then, using the distance to the fitted plane as a constraint, point clouds whose distance is less than the preset plane distance are extracted. The number of point cloud points in this predetermined plane point cloud is then determined. By comparing the number of point clouds, the predetermined plane point cloud with the most point clouds is determined to be the solid-state radar plane point cloud or the mechanical radar plane point cloud.

[0048] Optionally, the preset threshold number of iterations can be selected as 300, and the plane preset distance can be selected as 0.05 meters. It should be noted that the selection of the above specific values ​​can be appropriately determined according to actual conditions. The above values ​​are only used as preferred examples and do not limit the scope of protection of this application.

[0049] exist Figure 1 In the illustrated embodiment, the extrinsic parameter calibration method of the present application includes process S103 , performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to determine the center position of the measurement circular hole.

[0050] In this embodiment, after obtaining the solid-state radar plane point cloud and the mechanical radar plane point cloud, it is necessary to further detect the circular hole in the plane point cloud to obtain the edge of the circular hole and then determine the center position of the circular hole.

[0051] Optionally, circular hole edge detection is performed on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to determine the measured circular hole center position, including: performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to obtain corresponding circular hole edge point clouds; projecting the circular hole edge point cloud onto the plane where the circular hole calibration plate is located to obtain two-dimensional circular hole edge points, and then obtaining the measured circular hole center position.

[0052] In this optional embodiment, circular hole edge features are used to perform circular hole edge detection on both the solid-state radar plane point cloud and the mechanical radar plane point cloud, obtaining corresponding circular hole edge point clouds. The resulting circular hole edge point clouds are then projected onto the plane of the circular hole calibration plate, converting the three-dimensional point cloud data into two-dimensional data to obtain two-dimensional circular hole edge points. This provides a rough circular hole edge on the two-dimensional plane, and the circular hole center position is then determined based on the circular hole edge.

[0053] Optionally, circular hole edge detection is performed on the mechanical radar plane point cloud respectively to obtain the corresponding circular hole edge point cloud, including: storing the distance information of each point cloud point in the mechanical radar plane point cloud in the form of a wire harness; judging whether the distance between the point cloud point corresponding to each wire harness and its left and right field points exceeds a preset distance threshold; if the distance is greater than the preset distance threshold and the point cloud point meets the global edge point feature, the point cloud point is determined as an edge point.

[0054] In this optional embodiment, the distance information of each point cloud point in the mechanical radar plane point cloud is stored in advance in the form of a wire bundle, and it is determined whether the distance between the point cloud point corresponding to each wire bundle and its left and right field points exceeds a preset distance threshold. If the distance is greater than the preset distance threshold and the corresponding point cloud point meets the global edge point feature, the point cloud point is determined as an edge point.

[0055] Specifically, a multi-line rotating laser radar is used to scan the point cloud of the circular hole calibration plate. For example, 16 lines, 32 lines, 64 lines, etc. This type of radar has multiple laser transmitters and receivers in the vertical direction. Through the rotation of the motor, multiple beams are obtained. The more beams there are, the more complete the surface contour of the detected object is. The angle formed between each connected laser transmitter is its angular resolution. Generally, the angles are the same, but there are also different angles. When the mechanical radar collects the power information of the circular hole calibration plate, the laser beam is divided, and then the laser points in each line are processed separately. The distance between the current point and its left neighboring point and its right neighboring point in the laser coordinate system is compared. If the difference in distance between the point and the left point or the right point is greater than the preset distance, for example, 0.1m, the point is determined to be an edge point. Edge points are extracted for all laser beams using this method, and all global edge points are finally obtained. Because, for the circular hole calibration plate, the edge points include the edge of the calibration plate and the edge of the circular hole. Since this application mainly processes the circular hole, the area of ​​interest corresponding to the circular hole is extracted from the global edge point to obtain the edge point corresponding to the circular hole.

[0056] Specifically, when detecting the circular hole edge of the solid-state radar plane point cloud, with the help of the PCL (Point Cloud Library) edge detection tool, the segmented plane points are first used to establish a spatial topological relationship between the scattered data using KD-tree, and based on this structure, the k nearest neighbor points of the sampling point are obtained as local surface reference data. The micro-cut plane of the data is fitted by least squares and projected onto the micro-cut plane. The edge features of the circular hole calibration plate point cloud are identified based on the maximum angle between the sampling point and the projection points corresponding to its k nearest neighbors (the value of this algorithm is π / 2), and finally the edge points are determined.

[0057] Optionally, the circular hole edge point cloud is projected onto the plane where the circular hole calibration plate is located to obtain two-dimensional circular hole edge points, and then the measured circular hole center position is obtained, including: using a two-dimensional circle model, sampling a preset number of times in the two-dimensional circular hole edge points, and sampling three edge points each time, fitting a preset number of circular hole edge lines, wherein the three edge points are not sampled repeatedly; the two-dimensional circular hole edge points whose distance to the circular hole edge line is less than a preset distance threshold are determined as the circular hole inner points of the corresponding circular hole edge line; the number of circular hole inner points of each circular hole edge line is determined, and the circular hole edge line with the most circular hole inner points is determined as the circular hole edge line of the circular hole calibration plate, and then the measured circular hole center position is determined.

[0058] In this optional embodiment, after obtaining the two-dimensional circular hole edge points, it is necessary to further determine the edge line of the circular hole. First, a preset number of sampling times are performed on the two-dimensional circular hole edge points, and three edge points are obtained in each sampling. Then, the corresponding circular edge line is fitted based on these three edge points. Therefore, after multiple samplings, multiple circular hole edge lines can be obtained. The two-dimensional circular hole edge points whose distance from the circular hole edge line is less than the preset distance threshold are determined as the circular hole inner points of the corresponding circular hole edge line, and then the number of circular hole inner points corresponding to each circular hole edge line is calculated. By comparing the number of circular hole inner points, the circular hole edge line with the largest number of circular hole inner points is determined as the circular hole edge line of the circular hole calibration plate. Then, the center position of the circular hole is measured based on the circular hole edge line. Among them, the process of determining the center position of the circle based on the circumference is a common technology and will not be described in detail in this application.

[0059] Optionally, the preset distance threshold can be selected as 0.02 meters, wherein the specific value of the threshold can be appropriately determined according to actual conditions. The above values ​​are merely preferred examples and do not limit the scope of protection of this application.

[0060] exist Figure 1 In the illustrated embodiment, the external parameter calibration method of the present application includes process S104, verifying the measured circular hole center position according to the real circular hole center position in the circular hole calibration plate, and determining the circular hole position data that meets the verification conditions as the registration circular hole position data.

[0061] In this implementation, the position of the circular hole calibration plate is obtained by sensing and calculating the point cloud of the circular hole calibration plate. To ensure the accuracy of the circular hole calibration plate detection, it is necessary to verify the circular hole position based on the actual circular hole calibration plate. The position that meets the verification conditions is used as the final registration circular hole position data.

[0062] Optionally, the measured circular hole center position is verified according to the real circular hole center position in the circular hole calibration plate, and the one that meets the verification conditions is determined as the registration circular hole position data, including: determining the correspondence between the measured circular hole center position and the real circular hole center position; according to the correspondence, calculating the distance error between multiple measured circular hole center positions and the corresponding multiple real circular hole center positions; if the distance error is less than a preset threshold, the measured circular hole center position is determined as the registration circular hole position data.

[0063] In this optional embodiment, multiple circular holes are provided in a circular hole calibration plate. By scanning the circular hole calibration plate to obtain point cloud data, and performing the aforementioned calculations, the measured circular hole center positions are obtained. By calculating the distance error between the multiple measured circular hole center positions and the corresponding multiple true circular hole center positions, the magnitude of the error is used to determine whether the obtained measured circular hole center positions truly reflect the relationship between the true circular hole positions. If the distance error is less than a preset threshold, the measurement result is considered accurate, and the measured circular hole center positions are determined as the registered circular hole position data.

[0064] Optionally, the preset threshold value may be selected as 0.06 meters. It should be noted that the selection of the specific value of the threshold value may be appropriately determined according to actual conditions and does not limit the scope of protection of this application.

[0065] Optionally, determining the correspondence between the measured circular hole center position and the real circular hole center position includes: determining, in the radar coordinate system, the angle between the line connecting the center position of each measured circular hole and the coordinate origin and the xy plane; comparing each angle, wherein the measured circular hole center position corresponding to the larger angle corresponds to the real circular hole center position at the upper part of the circular hole calibration plate, and the measured circular hole center position corresponding to the smaller angle corresponds to the real circular hole center position at the lower part of the circular hole calibration plate.

[0066] In this optional embodiment, when performing precision calibration of the corresponding cloud circular hole position, the first step is to determine the correspondence between the measured circular hole center position and the true circular hole center position. In the radar coordinate system, the radar is used as the coordinate system origin, the radar's forward direction is the x-axis direction, the left direction is the y-axis direction, and the upward direction is the z-axis direction. In the radar coordinate system, the angle between the line connecting the center position of each measured circular hole and the coordinate origin and the xy plane is determined. A larger angle indicates a higher height from the ground, corresponding to the upper circular hole in the circular hole calibration plate; a smaller angle indicates a lower height from the ground, corresponding to the lower circular hole in the circular hole calibration plate.

[0067] Optionally, determining the correspondence between the measured circular hole center position and the real circular hole center position includes: determining the projection points of the center positions of each measured circular hole on the xy plane in the radar coordinate system; comparing the positions of the projection points, wherein the measured circular hole center position corresponding to the projection point on the left corresponds to the real circular hole center position on the left side of the circular hole calibration plate, and the measured circular hole center position corresponding to the projection point on the right corresponds to the real circular hole center position on the right side of the circular hole calibration plate.

[0068] In this optional embodiment, the center positions of each measurement circular hole are projected onto the xy plane to obtain projection points. Based on the relative positions of the projection points, the measurement circular hole center position corresponding to the projection point on the left side corresponds to the true circular hole center position on the left side of the circular hole calibration plate, and the measurement circular hole center position corresponding to the projection point on the right side corresponds to the true circular hole center position on the right side of the circular hole calibration plate.

[0069] exist Figure 1 In the illustrated embodiment, the extrinsic parameter calibration method of the present application includes process S105, which solves the extrinsic parameters by using multiple sets of registration circular hole position data to obtain the extrinsic parameters of the solid-state radar and the extrinsic parameters of the mechanical radar.

[0070] In this embodiment, after obtaining multiple sets of registration circular hole position data, common technical means in the prior art are used to calculate the external parameters of the solid-state radar and the mechanical radar respectively, and finally the corresponding external parameters of the solid-state radar and the mechanical radar are obtained.

[0071] Specifically, the extrinsic parameter solution process requires 3 to 4 sets of registered circular hole calibration plate data. The extrinsic parameter solution method adopts a combination of ICP and nonlinear optimization methods to align the coordinates of the circular hole under each radar system. The cost function is to minimize the three-dimensional distance between the registration points. The extrinsic parameters output by ICP are used as the initial values ​​of the nonlinear optimization solution. The ceres library is used for optimization and solution, and the final extrinsic parameter matrix is ​​output.

[0072] Specifically, Figure 2 An embodiment of the external parameter calibration method of the present application is shown.

[0073] like Figure 2 As shown in the figure, the original point cloud of the circular hole calibration plate is first acquired using solid-state radar and mechanical radar respectively. After preprocessing such as straight-through filtering, plane segmentation is performed to obtain the plane point cloud corresponding to the circular hole calibration plate in the original point cloud. Edge detection is then performed on the acquired plane point cloud to obtain the edge points corresponding to the circular hole in the calibration plate. The center of the circular hole is then determined based on the circular hole edge points. The detected circular hole center positions are then sorted and verified with the actual circular hole center positions. Finally, the coordinates of the detected circular hole center are determined. The coordinates of multiple sets of circular hole center are used to solve the external parameters of the solid-state radar and mechanical radar to obtain the final result.

[0074] The external parameter calibration method of the present application calibrates the solid-state radar and the mechanical radar through a circular hole calibration plate in the common viewing area of ​​the solid-state radar and the mechanical radar, and calibrates the external parameters by detecting the circular hole calibration plate, thereby improving the accuracy of the external parameter calibration and the reliability of multi-laser fusion.

[0075] Figure 3An embodiment of the external parameter calibration system of the present application is shown.

[0076] exist Figure 3 In the illustrated embodiment, the extrinsic parameter calibration system of the present application includes: a module 301 for scanning a circular hole calibration plate within a common viewing area using a solid-state radar and a mechanical radar, respectively, when the carrier is stationary, to obtain solid-state radar point cloud data and mechanical radar point cloud data, respectively;

[0077] Module 302 for performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data, respectively, to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate;

[0078] Module 303 for performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively, and determining the center position of the measured circular hole;

[0079] Module 304 for verifying the measured circular hole center position according to the real circular hole center position in the circular hole calibration plate, and determining the position that meets the verification conditions as the registration circular hole position data;

[0080] A module 305 is used to solve the external parameters by using multiple sets of registration circular hole position data to obtain the external parameters of the solid-state radar and the external parameters of the mechanical radar.

[0081] Optionally, in module 301, the circular hole calibration plate is scanned multiple times by the solid-state radar to obtain multiple frames of solid-state radar original point cloud data; the solid-state radar original point cloud data is straight-through filtered to extract the area of ​​interest corresponding to the circular hole calibration plate to obtain solid-state radar point cloud data.

[0082] Optionally, in module 301, the circular hole calibration plate is scanned once by the mechanical radar to obtain the mechanical radar original point cloud data; the area of ​​interest corresponding to the circular hole calibration plate is extracted from the mechanical radar original point cloud data to obtain the mechanical radar point cloud data.

[0083] Optionally, in module 302, three point cloud points with a preset threshold number of times are selected from the solid-state radar point cloud data or the mechanical radar point cloud data, and corresponding multiple fitting planes are constructed respectively; the point cloud data in the solid-state radar point cloud data or the mechanical radar point cloud data whose distance from the fitting plane is less than the preset distance of the plane is determined as a predetermined plane point cloud; the number of point cloud points in the multiple predetermined plane point clouds is determined, and the predetermined plane point cloud with the largest number of point cloud points is determined as the corresponding solid-state radar plane point cloud or the mechanical radar plane point cloud.

[0084] Optionally, in module 303, circular hole edge detection is performed on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to obtain corresponding circular hole edge point clouds; the circular hole edge point clouds are projected onto the plane where the circular hole calibration plate is located to obtain two-dimensional circular hole edge points, and then the measured circular hole center position is obtained.

[0085] Optionally, in module 303, the distance information of each point cloud point in the mechanical radar plane point cloud is stored in the form of a wire harness; it is determined whether the distance between the point cloud point corresponding to each wire harness and its left and right field points exceeds a preset distance threshold; if the distance is greater than the preset distance threshold and the point cloud point meets the global edge point characteristics, the point cloud point is determined as an edge point.

[0086] Optionally, in module 303, a preset number of sampling times are performed on the two-dimensional circular hole edge points using the two-dimensional circle model, with three edge points sampled each time, to fit a preset number of circular hole edge lines, where the three edge points are not sampled repeatedly. Two-dimensional circular hole edge points whose distance from the circular hole edge line is less than a preset distance threshold are determined as the corresponding circular hole in-hole points of the circular hole edge line. The number of circular hole in-hole points on each circular hole edge line is determined, and the circular hole edge line with the most circular hole in-hole points is determined as the circular hole edge line of the circular hole calibration plate, thereby determining the center position of the measured circular hole.

[0087] Optionally, in module 304, the distance error between the distances between the multiple measured circular hole center positions and the corresponding multiple real circular hole center positions is calculated. If the distance error is less than a preset threshold, the measured circular hole center position is determined as the registration circular hole position data.

[0088] Optionally, in module 304, in the radar coordinate system, the angle between the line connecting the center position of each measuring circular hole and the coordinate origin and the xy plane is determined; and each angle is compared, where the center position of the measuring circular hole corresponding to the larger angle corresponds to the center position of the actual circular hole at the upper part of the circular hole calibration plate, and the center position of the measuring circular hole corresponding to the smaller angle corresponds to the center position of the actual circular hole at the lower part of the circular hole calibration plate.

[0089] Optionally, in module 304, in the radar coordinate system, the projection points of the center positions of each measuring circular hole on the xy plane are determined; the positions of the projection points are compared, where the center position of the measuring circular hole corresponding to the projection point on the left corresponds to the center position of the actual circular hole on the left side of the circular hole calibration plate, and the center position of the measuring circular hole corresponding to the projection point on the right corresponds to the center position of the actual circular hole on the right side of the circular hole calibration plate.

[0090] The external parameter calibration system of the present application calibrates the solid-state radar and the mechanical radar through a circular hole calibration plate in their common viewing area, and calibrates the external parameters by detecting the circular hole calibration plate, thereby improving the accuracy of the external parameter calibration and the reliability of multi-laser fusion.

[0091] The present application also provides a computer-readable storage medium storing computer instructions, wherein the computer instructions are operated to perform the extrinsic parameter calibration method described in any embodiment. The storage medium can be directly in hardware, in a software module executed by a processor, or in a combination of the two.

[0092] The software modules may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such that the processor can read information from and write information to the storage medium.

[0093] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration. In the alternative, the storage medium may be integral to the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In the alternative, the processor and storage medium may reside as discrete components in the user terminal.

[0094] An embodiment of the present application also provides a computer device, which includes a processor and a memory, wherein the memory stores computer instructions, wherein: the processor operates the computer instructions to execute the external parameter calibration method described in any embodiment.

[0095] In the embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0096] 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.

[0097] The above are merely embodiments of the present application and are not intended to limit the patent scope of the present application. Any equivalent structural transformations made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for calibrating an external parameter, characterized in that: include: When the carrier is stationary, the solid-state radar and the mechanical radar scan the circular hole calibration plate in the common view area respectively to obtain solid-state radar point cloud data and mechanical radar point cloud data respectively; performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data respectively, and determining the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate; Performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to determine the center position of the measurement circular hole; Verifying the center position of the measured circular hole according to the center position of the real circular hole in the circular hole calibration plate, and determining the position that meets the verification conditions as the registration circular hole position data; The external parameters are solved by using multiple sets of the registration circular hole position data to obtain the external parameters of the solid-state radar and the external parameters of the mechanical radar.

2. The external parameter calibration method according to claim 1, characterized in that The solid-state radar and the mechanical radar are used to scan the circular hole calibration plate in the common view area respectively, and the solid-state radar point cloud data and the mechanical radar point cloud data are obtained respectively, including: Scanning the circular hole calibration plate multiple times by the solid-state radar obtains multiple frames of solid-state radar original point cloud data; The solid-state radar original point cloud data is subjected to straight-through filtering, and the region of interest corresponding to the circular hole calibration plate is extracted to obtain the solid-state radar point cloud data.

3. The external parameter calibration method according to claim 1, characterized in that The method further comprises: scanning the circular hole calibration plate in the common view area by the solid-state radar and the mechanical radar respectively to obtain solid-state radar point cloud data and mechanical radar point cloud data respectively; Scanning the circular hole calibration plate once by the mechanical radar obtains the original point cloud data of the mechanical radar; The region of interest corresponding to the circular hole calibration plate is extracted from the original point cloud data of the mechanical radar to obtain the point cloud data of the mechanical radar.

4. The external parameter calibration method according to claim 1, characterized in that The step of segmenting the solid-state radar point cloud data and the mechanical radar point cloud data to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate includes: Selecting three point cloud points with a preset threshold number of times from the solid-state radar point cloud data or the mechanical radar point cloud data, and constructing corresponding multiple fitting planes respectively; Determining, in the solid-state radar point cloud data or the mechanical radar point cloud data, point cloud data whose distance from the fitting plane is less than a preset distance from the plane as a predetermined plane point cloud; The number of point cloud points in the plurality of predetermined planar point clouds is determined, and the predetermined planar point cloud with the largest number of point cloud points is determined as the corresponding solid-state radar planar point cloud or the mechanical radar planar point cloud.

5. The external parameter calibration method according to claim 1, characterized in that: The performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to determine the center position of the measuring circular hole includes: Performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud respectively to obtain corresponding circular hole edge point clouds; The circular hole edge point cloud is projected onto the plane where the circular hole calibration plate is located to obtain a two-dimensional circular hole edge point, and then the center position of the measured circular hole is obtained.

6. The external parameter calibration method according to claim 5, characterized in that: The performing circular hole edge detection on the mechanical radar plane point cloud to obtain corresponding circular hole edge point clouds includes: Storing the distance information of each point cloud point in the mechanical radar plane point cloud by means of a wire harness; Determine whether the distance between the point cloud point corresponding to each harness and its left and right field points exceeds the preset distance threshold; If the distance is greater than the preset distance threshold and the point cloud point meets the global edge point feature, the point cloud point is determined as an edge point.

7. The external parameter calibration method according to claim 5, characterized in that: The step of projecting the circular hole edge point cloud onto the plane where the circular hole calibration plate is located to obtain a two-dimensional circular hole edge point, and then obtaining the measured circular hole center position, includes: Using a two-dimensional circle model, sampling is performed a preset number of times at the edge points of the two-dimensional circular hole, and three edge points are sampled each time, so as to fit a preset number of circular hole edge lines, wherein the three edge points are not sampled repeatedly; Determine the two-dimensional circular hole edge point whose distance from the circular hole edge line is less than a preset distance threshold as the corresponding circular hole inner point of the circular hole edge line; The number of circular hole inner points of each circular hole edge line is determined, and the circular hole edge line with the most circular hole inner points is determined as the circular hole edge line of the circular hole calibration plate, thereby determining the center position of the measurement circular hole.

8. The external parameter calibration method according to claim 1, characterized in that: The method of verifying the measured circular hole center position according to the real circular hole center position in the circular hole calibration plate and determining the position that meets the verification conditions as the registration circular hole position data includes: Determine the correspondence between the center position of the measured circular hole and the center position of the real circular hole; According to the correspondence, the distance error between the distances between the multiple measured circular hole center positions and the corresponding multiple real circular hole center positions is calculated. If the distance error is less than a preset threshold, the measured circular hole center position is determined as the registration circular hole position data.

9. The external parameter calibration method according to claim 8, characterized in that: Determining the correspondence between the center position of the measured circular hole and the center position of the real circular hole includes: In the radar coordinate system, determining the angle between the line connecting the center position of each measuring circular hole and the coordinate origin and the xy plane; A comparison is performed between the angles, wherein the center position of the measuring circular hole corresponding to the larger angle corresponds to the center position of the real circular hole at the top of the circular hole calibration plate, and the center position of the measuring circular hole corresponding to the smaller angle corresponds to the center position of the real circular hole at the bottom of the circular hole calibration plate.

10. The external parameter calibration method according to claim 8, characterized in that: Determining the correspondence between the center position of the measured circular hole and the center position of the real circular hole includes: In the radar coordinate system, determining the projection point of the center position of each measuring circular hole on the xy plane; The positions of the projection points are compared, wherein the center position of the measuring circular hole corresponding to the projection point on the left corresponds to the center position of the real circular hole on the left side of the circular hole calibration plate, and the center position of the measuring circular hole corresponding to the projection point on the right corresponds to the center position of the real circular hole on the right side of the circular hole calibration plate.

11. An external parameter calibration system, characterized in that: include: A module for scanning the circular hole calibration plate in the common viewing area by the solid-state radar and the mechanical radar respectively when the carrier is stationary, to obtain the solid-state radar point cloud data and the mechanical radar point cloud data respectively; A module for performing plane segmentation on the solid-state radar point cloud data and the mechanical radar point cloud data, respectively, to determine the solid-state radar plane point cloud and the mechanical radar plane point cloud corresponding to the circular hole calibration plate; A module for performing circular hole edge detection on the solid-state radar plane point cloud and the mechanical radar plane point cloud, respectively, and determining the center position of the measured circular hole; A module for verifying the center position of the measured circular hole according to the center position of the real circular hole in the circular hole calibration plate, and determining the position that meets the verification conditions as the registration circular hole position data; A module for solving external parameters using multiple sets of the registration circular hole position data to obtain corresponding external parameters of the solid-state radar and the mechanical radar.

Images