A road boundary extraction method based on multi-line laser radar

Abstract

In view of the limitations of the prior art, the present invention proposes a method for extracting road boundary lines using a multi-line laser radar. While ensuring accuracy, the present invention greatly reduces the cost of the hardware system, and is convenient for popularization and updating of high-precision maps; the present invention has more accurate spatial information and can achieve practical application accuracy; the present invention is not based on turning point features, but extracts boundary points through ground features, distance features, and straight line features, eliminating interference caused by noise points on the road and roadside buildings, trees, and grass, and accurately extracts road boundary lines; the present invention directly processes points and is not affected by the resolution of the point cloud. In addition, the present invention fits the curve area through multi-frame fusion and ROI adjustment, and can accurately extract the road boundary of the curve area; at the same time, multi-frame fusion can supplement the road boundary information, and can accurately and completely extract various types of road boundaries.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving and automatic construction of high-precision maps, and specifically to the application of multi-line radar in the above technical field, and more specifically to a method for extracting road boundary lines from a multi-line laser radar. Background Art

[0002] Road boundary lines are the basic elements of high-precision maps. How to accurately and completely extract road boundary lines is the key to automatically constructing high-precision maps. At the same time, road boundary lines are also important local information for autonomous driving after losing positioning signals. Relying on road boundary information, it can avoid driving out of the lane. Multi-line radar has accurate spatial information, but due to the low point cloud density, the development of detection technology is not yet perfect. At present, the main problems faced by using multi-line lidar to extract road boundary lines include the following: First, there is interference on the road. When a vehicle is driving on the road, there are usually other vehicles on the road. Even if there are only a few vehicles on the road, the vehicle is still easily identified as a road boundary. Moreover, the vehicle will block the road boundary. In addition to the vehicle point cloud, the road surface point cloud also affects the extraction of road boundary lines; second, there are many interferences on the road side. The point cloud features of the connection between the road surface and the roadside buildings, trees and pedestrians are similar to the road boundary features, and are easily identified as road boundaries.

[0003] The multi-line radar based methods can be divided into line separation methods and non-line separation methods according to the method of processing point clouds. The non-line separation method mainly processes through density after voxelization, and the line separation methods mainly include methods based on turning point features, methods based on elevation difference, and methods based on line fitting, etc.

[0004] Specifically, the voxel-based method has the advantages of fast processing speed and accurate classification, but due to the sparsity of point clouds, it is difficult to achieve satisfactory results: too small a grid will lead to an increase in the number of empty grids, slowing down the processing speed; too large a grid will cause the boundary grid to contain road point clouds, affecting the accuracy of road boundary line extraction. As shown in the Chinese invention application published on May 14, 2019: A road edge detection method based on laser point clouds, it belongs to a method based on turning point features; the method based on turning point features after line division is the mainstream method now because of its clear features and fast calculation, but the existing technology in this regard is easily interfered by roadside turning points and noise points: First, there is interference in the road. When a vehicle is driving on the road, there are usually other vehicles on the road. Even if there are only a few vehicles on the road, the vehicle is still easily identified as a road boundary, and the vehicle will block the road boundary. In addition to the vehicle point cloud, the road surface point cloud also affects the extraction of the road boundary line; second, there are many interferences on the roadside. The point cloud features of the connection between the road surface and the roadside buildings, trees and pedestrians are similar to the road boundary features and are easily identified as road boundaries. Therefore, the existing technology still has certain limitations. Summary of the invention

[0005] In view of the limitations of the prior art, the present invention proposes a method for extracting road boundary lines using a multi-line laser radar. The technical solution adopted by the present invention is:

[0006] A method for extracting road boundary lines using a multi-line laser radar comprises the following steps:

[0007] S1, obtaining input radar data, identifying and removing obstacles in the radar data;

[0008] S2, setting the region of interest in a preset vehicle local coordinate system;

[0009] S3, extracting a road point cloud with shoulder content and ground content from the radar data according to the region of interest and at a preset fitting plane threshold distance;

[0010] S4, filtering the road point cloud by lowering the fitting plane threshold distance to remove ground content in the road point cloud;

[0011] S5, extracting candidate boundary points from the results filtered in step S4; performing boundary line fitting based on the candidate boundary points to extract the road boundary line of the current frame;

[0012] S6, determining whether a preset number of road boundary lines have been extracted; if not, executing step S7; if yes, executing step S8;

[0013] S7, calculating the vehicle speed of the current frame through the odometer information in the radar data; returning to step S2 to adjust the region of interest based on the vehicle speed of the current frame, and performing a road boundary line extraction process for the next frame;

[0014] S8, performing straight line fitting on the road boundary lines extracted from each frame, superimposing the boundary point clouds obtained by the straight line fitting, and completing the extraction of the road boundary lines from the radar data.

[0015] Compared with the MLS-based method, the present invention greatly reduces the cost of the hardware system while ensuring accuracy, and is convenient for popularization and updating of high-precision maps; compared with the image-based method, the present invention has more accurate spatial information and can achieve practical application accuracy; compared with the method based on turning points after line division, the present invention is not based on turning point features, but extracts boundary points through ground features, distance features and straight line features, eliminating the interference caused by noise points in the road and roadside buildings, trees and grass, and accurately extracts road boundary lines; compared with the voxel-based method, the present invention directly processes points and is not affected by the resolution of the point cloud. In addition, the present invention fits the curve area through multi-frame fusion and ROI adjustment, and can accurately extract the road boundary of the curve area; at the same time, multi-frame fusion can supplement the road boundary information, and can accurately and completely extract various types of road boundaries.

[0016] As a preferred solution, for a vehicle local coordinate system with the vehicle head direction as the X-axis, the vehicle side direction as the Y-axis, and the vehicle vertical direction as the Z-axis, the region of interest is set in step S2 according to the following formula:

[0017] P ROI ={p i (x i ,y i , z i )|-x limit <x i <x limit , -y limit <y i <y limit};

[0018] Among them, P ROI represents the set of points in the region of interest; p i Represents the point cloud of radar data; (x i ,y i , z i ) represents the point cloud p i The coordinates of limit Indicates the range of the X axis; y limit Indicates the range of the Y axis.

[0019] Furthermore, the x limit Set it according to the following formula:

[0020]

[0021] Among them, v t Indicates the vehicle speed.

[0022] As a preferred solution, in step S5, the candidate boundary points are extracted by the following method: the result P after filtering in step S5 is divided into the left and right points according to the straight line where the vehicle head is facing. onboundary Divided into P left and P right ; P left and P right It is divided into several parts along the direction of the front of the car, starting from P left and P right The points closest to the vehicle in each divided part are selected as candidate boundary points.

[0023] As a preferred solution, in step S4, the road point cloud is filtered through the following process:

[0024] S41, lowering the fitting plane threshold distance, performing plane fitting on the remaining point clouds in the road point cloud, and removing the plane point clouds obtained by plane fitting;

[0025] S42, determine whether the number of plane point clouds obtained by this round of plane fitting is less than a preset threshold value PNum limit ; If not, return to step S41; If so, output the remaining point cloud in the road point cloud as the result after filtering in step S4.

[0026] Furthermore, in steps S3, S4 and S5, a random sampling consistency algorithm is used to perform extraction and fitting.

[0027] Furthermore, in step S1, obstacles in the radar data are removed by a deep learning-based method.

[0028] The present invention also provides the following contents:

[0029] A multi-line laser radar road boundary line extraction system comprises a radar data acquisition module, an area of ​​interest setting module, a point cloud rough extraction module, a point cloud fine filtering module, a boundary line fitting module, a frame number judgment module, a speed calculation module and a multi-frame optimization module; the area of ​​interest setting module is connected to the radar data acquisition module, the point cloud rough extraction module and the speed calculation module; the point cloud fine filtering module is connected to the point cloud rough extraction module and the boundary line fitting module; the frame number judgment module is connected to the boundary line fitting module, the speed calculation module and the multi-frame optimization module; wherein:

[0030] The radar data acquisition module is used to acquire input radar data and identify and remove obstacles in the radar data;

[0031] The region of interest setting module is used to set the region of interest in a preset vehicle local coordinate system;

[0032] The point cloud rough extraction module is used to extract the road point cloud with shoulder content and ground content from the radar data according to the region of interest and at a preset fitting plane threshold distance;

[0033] The point cloud fine filtering module is used to filter the road point cloud by lowering the fitting plane threshold distance to remove ground content in the road point cloud;

[0034] The boundary line fitting module is used to extract candidate boundary points from the results filtered by the point cloud fine filtering module; perform boundary line fitting based on the candidate boundary points to extract the road boundary line of the current frame;

[0035] The frame number determination module is used to determine whether the road boundary line of the preset frame number has been extracted; if not, turn to the speed calculation module; if so, turn to the multi-frame optimization module;

[0036] The speed calculation module is used to calculate the vehicle speed of the current frame through the odometer information in the radar data; according to the vehicle speed of the current frame, the interest region setting module is turned to adjust the interest region and perform the road boundary line extraction process of the next frame;

[0037] The multi-frame optimization module is used to perform straight line fitting on the road boundary lines extracted from each frame, and to superimpose the boundary point clouds obtained by the straight line fitting to complete the extraction of the road boundary lines from the radar data.

[0038] A storage medium stores a computer program, which, when executed by a processor, implements the steps of the aforementioned multi-line laser radar road boundary line extraction method.

[0039] A computer device includes a storage medium, a processor, and a computer program stored in the storage medium and executable by the processor. When the computer program is executed by the processor, the steps of the aforementioned multi-line laser radar road boundary line extraction method are performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of the steps of the multi-line laser radar road boundary line extraction method provided in Example 1 of the present invention;

[0041] Figure 2 This is an example of the road boundary point cloud result extracted in Example 1 of the present invention;

[0042] Figure 3 Schematic diagram of the multi-line laser radar road boundary line extraction system provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0043] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0044] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present application.

[0045] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. The singular forms of "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0046] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not have to be used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0047] In addition, in the description of this application, unless otherwise specified, "multiple" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0048] In order to solve the limitations of the prior art, this embodiment provides a technical solution, and the technical solution of the present invention is further described below in conjunction with the drawings and embodiments.

[0049] Example 1

[0050] Please refer to Figure 1 , a multi-line laser radar road boundary line extraction method, comprising the following steps:

[0051] S1, obtaining input radar data, identifying and removing obstacles in the radar data;

[0052] S2, setting the region of interest in a preset vehicle local coordinate system;

[0053] S3, extracting a road point cloud with shoulder content and ground content from the radar data according to the region of interest and at a preset fitting plane threshold distance;

[0054] S4, filtering the road point cloud by lowering the fitting plane threshold distance to remove ground content in the road point cloud;

[0055] S5, extracting candidate boundary points from the results filtered in step S4; performing boundary line fitting based on the candidate boundary points to extract the road boundary line of the current frame;

[0056] S6, determining whether a preset number of road boundary lines have been extracted; if not, executing step S7; if yes, executing step S8;

[0057] S7, calculating the vehicle speed of the current frame through the odometer information in the radar data; returning to step S2 to adjust the region of interest based on the vehicle speed of the current frame, and performing a road boundary line extraction process for the next frame;

[0058] S8, performing straight line fitting on the road boundary lines extracted from each frame, superimposing the boundary point clouds obtained by the straight line fitting, and completing the extraction of the road boundary lines from the radar data.

[0059] Compared with the MLS-based method, the present invention greatly reduces the cost of the hardware system while ensuring accuracy, and is convenient for popularization and updating of high-precision maps; compared with the image-based method, the present invention has more accurate spatial information and can achieve practical application accuracy; compared with the method based on turning points after line division, the present invention is not based on turning point features, but extracts boundary points through ground features, distance features and straight line features, eliminating the interference caused by noise points in the road and roadside buildings, trees and grass, and accurately extracts road boundary lines; compared with the voxel-based method, the present invention directly processes points and is not affected by the resolution of the point cloud. In addition, the present invention fits the curve area through multi-frame fusion and ROI adjustment, and can accurately extract the road boundary of the curve area; at the same time, multi-frame fusion can supplement the road boundary information, and can accurately and completely extract various types of road boundaries.

[0060] Specifically, the present invention uses a method of directly fitting a plane to the point cloud to filter out the ground, solving the problem that the line division method is easily affected by inaccurate line division and noise points, and the voxel-based method is not accurate enough; multiple planes are used to fit the ground, solving the problem of incomplete traditional single plane fitting; a method of approximating a curve with a short straight line is used to extract the boundary line of the curved road, solving the problem of inaccurate extraction of the road boundary line in the curved area; and multi-frame optimization is used to reduce random errors when the algorithm makes mistakes. Compared with the existing method of extracting road boundary lines using multi-line laser radar, the present invention can accurately and completely extract road boundary lines and construct a road boundary point cloud map.

[0061] Furthermore, in step S1, obstacles in the radar data are removed by a deep learning-based method, such as PointPillars, Pixor and other methods.

[0062] As a preferred embodiment, for a vehicle local coordinate system with the vehicle head direction as the X-axis, the vehicle side direction as the Y-axis, and the vehicle vertical direction as the Z-axis, the region of interest is set in step S2 according to the following formula:

[0063] P ROI ={p i (x i ,y i , z i )|-x limit <x i <x limit , -y limit <y i <y limit};

[0064] Among them, P ROI represents the set of points in the region of interest; p i Represents the point cloud of radar data; (x i ,y i , z i ) represents the point cloud p i The coordinates of limit Indicates the range of the X axis; y limit Indicates the range of the Y axis.

[0065] Specifically, x limit and limit It can be determined according to the radar type and vehicle speed: the more radar beams and the slower the speed, the larger the ROI area; the faster the speed, the larger the x limit The smaller it is, the more you can fit the curve with a short straight line.

[0066] More specifically, the x limit Set it according to the following formula:

[0067]

[0068] Among them, v t Indicates the vehicle speed.

[0069] Furthermore, in steps S3, S4 and S5, a random sampling consensus (RANSAC) algorithm is used to perform extraction and fitting.

[0070] Specifically, in step S3, the fitting plane threshold distance is set to a larger value, such as 0.3 m, so that the shoulder and the ground are extracted together, and other point clouds are allowed.

[0071] In step S4, the fitting plane threshold distance can be adjusted to a smaller value, such as 0.03 m, and the ground is filtered out through multiple RANSAC operations to remove the ground point cloud. The remaining point cloud is filtered out using P onboundary express.

[0072] Specifically, as a preferred embodiment, in step S4, the road point cloud is filtered through the following process:

[0073] S41, lowering the fitting plane threshold distance, performing plane fitting on the remaining point clouds in the road point cloud, and removing the plane point clouds obtained by plane fitting;

[0074] S42, determine whether the number of plane point clouds obtained by this round of plane fitting is less than a preset threshold value PNum limit ; If not, return to step S41; If so, output the remaining point cloud in the road point cloud as the result after filtering in step S4.

[0075] Specifically, in an optional embodiment, PNum limit =15000.

[0076] As a preferred embodiment, in step S5, the candidate boundary points are extracted by the following method: the result P after filtering in step S5 is divided into the left and right points according to the straight line where the vehicle head is facing. onboundary Divided into P left and P righ t;P left and P right The vehicle is divided into N parts along the direction of the vehicle head (as an optional embodiment, N=600), starting from P left and P right The points closest to the vehicle in each divided part are selected as candidate boundary points.

[0077] The above process can be expressed as follows:

[0078]

[0079]

[0080]

[0081]

[0082] Where Δd is the width of each part along the front of the vehicle, ID is the code of each part, Indicates rounding down. Represents the set of candidate boundary points on the left, and uses the same method to extract candidate boundary points on the right

[0083] In step S5, RANSAC is used to and When fitting a straight line in the projection of the XY plane, the straight line threshold ε can be set very small: ε = 0.03. After the road boundary line is extracted, the road boundary line of the current frame can be transformed from the vehicle local coordinate system to the preset global coordinate system through the local to global transformation matrix; the transformation matrix can be obtained by registering the radar data after removing obstacles using the SLAM method.

[0084] In step S7, the speed is calculated using the odometer information according to the following formula:

[0085]

[0086] The odometer k Indicates t k Odometer information at the moment, v k Indicates t k The speed calculated at that moment.

[0087] After multiple frames of optimization, the final result can be found in Figure 2 It can be seen that the solution provided in this embodiment can accurately extract the road boundary of the curved area, can supplement the road boundary information, and can accurately and completely extract various types of road boundaries.

[0088] Example 2

[0089] A multi-line laser radar road boundary extraction system, see Figure 3, including a radar data acquisition module 1, an area of ​​interest setting module 2, a point cloud rough extraction module 3, a point cloud fine filtering module 4, a boundary line fitting module 5, a frame number judgment module 6, a speed calculation module 7 and a multi-frame optimization module 8; the area of ​​interest setting module 2 is connected to the radar data acquisition module 1, the point cloud rough extraction module 3 and the speed calculation module 7; the point cloud fine filtering module 4 is connected to the point cloud rough extraction module 3 and the boundary line fitting module 5; the frame number judgment module 6 is connected to the boundary line fitting module 5, the speed calculation module 7 and the multi-frame optimization module 8; wherein:

[0090] The radar data acquisition module 1 is used to acquire input radar data, identify and remove obstacles in the radar data;

[0091] The region of interest setting module 2 is used to set the region of interest in a preset vehicle local coordinate system;

[0092] The point cloud rough extraction module 3 is used to extract the road point cloud with shoulder content and ground content from the radar data according to the area of ​​interest and at a preset fitting plane threshold distance;

[0093] The point cloud fine filtering module 4 is used to filter the road point cloud by lowering the fitting plane threshold distance to remove the ground content in the road point cloud;

[0094] The boundary line fitting module 5 is used to extract candidate boundary points from the results filtered by the point cloud fine filtering module 4; perform boundary line fitting based on the candidate boundary points to extract the road boundary line of the current frame;

[0095] The frame number determination module 6 is used to determine whether the road boundary line of the preset frame number has been extracted; if not, turn to the speed calculation module 7; if so, turn to the multi-frame optimization module 8;

[0096] The speed calculation module 7 is used to calculate the vehicle speed of the current frame through the odometer information in the radar data; according to the vehicle speed of the current frame, the interest region setting module 2 is turned to adjust the interest region and perform the road boundary line extraction process of the next frame;

[0097] The multi-frame optimization module 8 is used to perform straight line fitting on the road boundary lines extracted from each frame, and to superimpose the boundary point clouds obtained by the straight line fitting to complete the extraction of the road boundary lines from the radar data.

[0098] Example 3

[0099] A storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the multi-line laser radar road boundary line extraction method in Example 1.

[0100] Example 4

[0101] A computer device includes a storage medium, a processor, and a computer program stored in the medium and executable by the processor. When the computer program is executed by the processor, the steps of the multi-line laser radar road boundary line extraction method in Example 1 are implemented.

[0102] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A multi-line laser radar road boundary line extraction method, characterized in that: The following steps are involved: S1, obtaining input radar data, identifying and removing obstacles in the radar data; S2, setting the region of interest in a preset vehicle local coordinate system; For the vehicle local coordinate system with the vehicle head direction as the X-axis, the vehicle side direction as the Y-axis, and the vehicle vertical direction as the Z-axis, the region of interest is set according to the following formula: ； in, represents a collection of points in the region of interest; A point cloud representing radar data; Representing point clouds The coordinates of Indicates the range of the X axis; Indicates the range of the Y axis; Set it according to the following formula: ； in, Indicates vehicle speed; S3, extracting a road point cloud with shoulder content and ground content from the radar data according to the region of interest and at a preset fitting plane threshold distance; S4, filtering the road point cloud by lowering the fitting plane threshold distance to remove ground content in the road point cloud; S5, extracting candidate boundary points from the results filtered in step S4; performing boundary line fitting based on the candidate boundary points to extract the road boundary line of the current frame; S6, determining whether a preset number of road boundary lines have been extracted; if not, executing step S7; if yes, executing step S8; S7, calculating the vehicle speed of the current frame through the odometer information in the radar data; returning to step S2 to adjust the region of interest based on the vehicle speed of the current frame, and performing a road boundary line extraction process for the next frame; S8, performing straight line fitting on the road boundary lines extracted from each frame, superimposing the boundary point clouds obtained by the straight line fitting, and completing the extraction of the road boundary lines from the radar data.

2. The multi-line laser radar road boundary line extraction method according to claim 1, characterized in that: In step S5, candidate boundary points are extracted by: filtering the result of step S5 left and right according to the straight line where the vehicle head is facing Divided into and ;Will and It is divided into several parts along the direction of the front of the car. and The points closest to the vehicle in each divided part are selected as candidate boundary points.

3. The multi-line laser radar road boundary line extraction method according to claim 1, characterized in that: In step S4, the road point cloud is filtered by the following process: S41, lowering the fitting plane threshold distance, performing plane fitting on the remaining point clouds in the road point cloud, and removing the plane point clouds obtained by plane fitting; S42, determining whether the number of plane point clouds obtained by plane fitting in this round is less than a preset threshold ; If not, return to step S41; If yes, the remaining point clouds in the road point cloud are output as the filtered result in step S4.

4. The multi-line laser radar road boundary line extraction method according to any one of claims 1 to 3, characterized in that: In the steps S3, S4 and S5, a random sampling consistency algorithm is used to perform extraction and fitting.

5. The multi-line laser radar road boundary line extraction method according to any one of claims 1 to 3, characterized in that: In the step S1, obstacles in the radar data are removed by a deep learning-based method.

6. A multi-line laser radar road boundary line extraction system, applied to the multi-line laser radar road boundary line extraction method according to any one of claims 1 to 5, characterized in that: The invention comprises a radar data acquisition module (1), an area of ​​interest setting module (2), a point cloud rough extraction module (3), a point cloud fine filtering module (4), a boundary line fitting module (5), a frame number judgment module (6), a speed calculation module (7) and a multi-frame optimization module (8); the area of ​​interest setting module (2) is connected to the radar data acquisition module (1), the point cloud rough extraction module (3) and the speed calculation module (7); the point cloud fine filtering module (4) is connected to the point cloud rough extraction module (3) and the boundary line fitting module (5); the frame number judgment module (6) is connected to the boundary line fitting module (5), the speed calculation module (7) and the multi-frame optimization module (8); wherein: The radar data acquisition module (1) is used to acquire input radar data and identify and remove obstacles in the radar data; The region of interest setting module (2) is used to set the region of interest using a preset vehicle local coordinate system; The point cloud rough extraction module (3) is used to extract a road point cloud with shoulder content and ground content from the radar data according to the region of interest and at a preset fitting plane threshold distance; The point cloud fine filtering module (4) is used to filter the road point cloud by lowering the fitting plane threshold distance to remove ground content in the road point cloud; The boundary line fitting module (5) is used to extract candidate boundary points from the results filtered by the point cloud fine filtering module (4); perform boundary line fitting based on the candidate boundary points to extract the road boundary line of the current frame; The frame number determination module (6) is used to determine whether a preset number of frames of road boundary lines have been extracted; if not, turning to the speed calculation module (7); if so, turning to the multi-frame optimization module (8); The speed calculation module (7) is used to calculate the vehicle speed of the current frame through the odometer information in the radar data; based on the vehicle speed of the current frame, the module turns to the region of interest setting module (2) to adjust the region of interest and perform a road boundary line extraction process for the next frame; The multi-frame optimization module (8) is used to perform straight line fitting on the road boundary lines extracted from each frame, and to superimpose the boundary point clouds obtained by the straight line fitting, thereby completing the extraction of the road boundary lines from the radar data.

7. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-line laser radar road boundary line extraction method as described in any one of claims 1 to 5 are implemented.

8. A computer device, characterized in that: It includes a storage medium, a processor, and a computer program stored in the storage medium and executable by the processor, and when the computer program is executed by the processor, the steps of the multi-line laser radar road boundary line extraction method as described in any one of claims 1 to 5 are implemented.

Images