A roadside detection method, device, terminal equipment and storage medium

Abstract

The present application relates to the field of environmental perception technology, and proposes a curb detection method, device, terminal equipment and storage medium. The curb detection method comprises: obtaining point cloud data of the road to be tested; performing rasterization processing on the point cloud data; for each grid obtained after the rasterization processing, computing the ground reference height value of the grid based on the point cloud contained in the grid, and filtering out the ground points in the point cloud contained in the grid based on the ground reference height value, and the difference between the height of the ground point and the ground reference height value is less than a first threshold; based on the point cloud contained in each grid after filtering out the ground point, detecting and obtaining the curb information of the road to be tested. The use of this curb detection method can improve the accuracy of curb detection.

Description

Technical Field

[0001] The present application relates to the field of environmental perception technology, and in particular to a curb detection method, apparatus, terminal device and storage medium. Background Art

[0002] The curb is the road boundary and is one of the important traffic elements in the driving environment of unmanned vehicles. It limits the passable area of ​​unmanned vehicles and is of great significance for the navigation, positioning, planning and control of unmanned vehicles during road driving.

[0003] At present, the conventional curb detection method usually uses equipment such as lidar to obtain point cloud data of the road and perform rasterization processing on the point cloud data; then, based on the characteristic that the curb is higher than the ground, the height difference between the grids is detected, and the grids whose height difference meets the set conditions are used as curb candidate areas, and curb marking points are detected from the curb candidate areas.

[0004] However, when the road surface is uneven, the above method may misidentify the uneven road surface area as a candidate curb area, resulting in a low accuracy rate of curb detection. Summary of the invention

[0005] In view of this, the embodiments of the present application provide a curb detection method, apparatus, terminal device and storage medium, which can improve the accuracy of curb detection.

[0006] A first aspect of an embodiment of the present application provides a road edge detection method, comprising:

[0007] Obtain point cloud data of the road to be tested;

[0008] Performing rasterization processing on the point cloud data;

[0009] For each grid obtained after the rasterization process, a ground reference height value of the grid is obtained according to the point cloud calculation contained in the grid, and ground points in the point cloud contained in the grid are filtered out according to the ground reference height value, and the difference between the height of the ground point and the ground reference height value is less than a first threshold value;

[0010] According to the point cloud contained in each grid after filtering out the ground points, the curb information of the road to be tested is detected and obtained.

[0011] In an embodiment of the present application, first, the point cloud data of the road to be tested is obtained, and the point cloud data is rasterized to obtain a plurality of grids; then, the ground reference height value corresponding to each grid is calculated based on the point cloud contained in each grid, and the points whose height difference between the point cloud contained in the grid and the corresponding ground reference height value is less than a set threshold are regarded as ground points for filtering out; finally, the curb information of the road to be tested is detected based on the point cloud data after filtering out the ground points. Assuming that a certain area of ​​the road is uneven, the ground reference height value corresponding to the grid where the area is located is an adaptation value obtained based on the point cloud calculation of the uneven road surface, which can more accurately represent the true ground height of the area, and thus can more accurately find the ground points in the point cloud of the area. Afterwards, by filtering out the ground points in the point cloud, the interference of the ground points on the curb detection can be removed, thereby improving the accuracy of the curb detection.

[0012] In an implementation of the embodiment of the present application, detecting and obtaining the curb information of the road to be tested according to the point cloud after filtering out the ground points contained in each grid may include:

[0013] Performing a clustering operation on the point cloud after filtering out the ground points contained in each of the grids to obtain an initial roadside area;

[0014] Determine a candidate roadside area according to the height difference between each grid contained in the initial roadside area; wherein the height of each grid is equal to the average value of the heights of all points of the point cloud contained therein;

[0015] Selecting a curb candidate point from the curb candidate area according to the height difference between each point included in the curb candidate area;

[0016] According to the preset roadside point determination conditions, a roadside marking point is selected from the roadside candidate points;

[0017] Curve fitting processing is performed on the roadside marking points to obtain a roadside marking curve of the road to be tested.

[0018] Furthermore, the initial roadside area includes M grids, and determining the roadside candidate area according to the height difference between the grids included in the initial roadside area may include:

[0019] Traversing every N adjacent grids in the M grids, if the heights of the N grids decrease or increase in sequence, and the absolute value of the height difference between the first grid and the last grid in the N grids is greater than a second threshold, then determining the N grids as the roadside candidate area, 2≤N<M;

[0020] The curb candidate area includes X points, and searching for curb candidate points in the curb candidate area according to the height differences between the points included in the curb candidate area may include:

[0021] Sorting the X points according to their arrangement positions in the roadside candidate area;

[0022] Select Y points in adjacent order from the X points, 3≤Y<X;

[0023] If the heights of the Y points decrease or increase successively, and the absolute value of the height difference between the first point and the last point among the Y points is greater than a third threshold, the midpoint of the Y points is determined as the curb candidate point of the curb candidate area.

[0024] Furthermore, selecting a roadside marking point from the roadside candidate points according to a preset roadside point determination condition may include:

[0025] For each of the roadside candidate points, if the difference between the height of the roadside candidate point and the ground reference height value of the grid where the roadside candidate point is located is greater than a fourth threshold, and the roadside candidate point is not within the target detection frame of the point cloud data, then the roadside candidate point is determined as a roadside marking point;

[0026] The target detection frame is generated and integrated into the point cloud data in the following manner:

[0027] Acquire a road image corresponding to the point cloud data;

[0028] Performing target detection processing on the road image for non-roadside objects to obtain the target detection frame;

[0029] The road image with the target detection frame and the point cloud data are fused.

[0030] Furthermore, performing curve fitting processing on the roadside marking points to obtain the roadside marking curve of the road to be tested may include:

[0031] Performing curve fitting processing on the roadside marking points to obtain the roadside curve of the road to be tested at the current moment;

[0032] If the difference between the curvature of the curb curve at the current moment and the curvature of the curb curve of the road to be tested at the previous moment is less than the fifth threshold, the curb curve at the current moment is determined as the curb marking curve of the road to be tested and output; wherein, the curb curve of the road to be tested at the previous moment is obtained by performing curb detection on the point cloud data of the road to be tested collected at the previous moment.

[0033] In an implementation of the embodiment of the present application, the obtaining the ground reference height value of the grid according to the point cloud computing contained in the grid may include:

[0034] Selecting a target point cloud corresponding to a partial road surface area of ​​the road to be tested from the point cloud data, wherein the absolute value of the height of each point included in the target point cloud is less than a sixth threshold;

[0035] Calculate the ground height empirical value according to the height of each point included in the target point cloud;

[0036] Calculate the ground height threshold of the grid according to the ground height empirical value;

[0037] Extracting valid points in the point cloud contained in the grid, wherein the height of the valid points is less than the ground height threshold;

[0038] Performing curve fitting processing on the effective points to obtain a target polynomial equation;

[0039] The target polynomial equation is solved to obtain the ground reference height value of the grid.

[0040] Furthermore, the performing rasterization processing on the point cloud data may specifically be:

[0041] Performing polar coordinate rasterization processing on the point cloud data;

[0042] The step of calculating the ground height threshold of the grid according to the ground height empirical value may include:

[0043] Dividing the grid into a plurality of grid partitions according to the radial distances from the polar coordinate origin;

[0044] For each of the grid partitions, the height relaxation amount corresponding to the grid partition is calculated based on the distribution range of the radial distance between the grid partition and the polar coordinate origin, and the sum of the ground height empirical value and the height relaxation amount is determined as the ground height threshold value of the grid partition.

[0045] A second aspect of an embodiment of the present application provides a road edge detection device, including:

[0046] A point cloud data acquisition module is used to acquire point cloud data of the road to be tested;

[0047] A rasterization processing module, used for performing rasterization processing on the point cloud data;

[0048] A ground point filtering module, for obtaining, for each grid obtained after the rasterization process, a ground reference height value of the grid according to the point cloud calculation contained in the grid, and filtering out ground points in the point cloud contained in the grid according to the ground reference height value, wherein the difference between the height of the ground point and the ground reference height value is less than a first threshold value;

[0049] The curb detection module is used to detect and obtain the curb information of the road to be tested based on the point cloud contained in each grid after filtering out the ground points.

[0050] The third aspect of an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the curb detection method provided in the first aspect of the embodiment of the present application is implemented.

[0051] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the curb detection method provided in the first aspect of the embodiment of the present application.

[0052] A fifth aspect of the embodiments of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device executes the curb detection method as described in the first aspect of the embodiments of the present application.

[0053] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flow chart of a roadside detection method provided in an embodiment of the present application;

[0055] Figure 2 It is a schematic diagram of the grid obtained after the point cloud data is processed by polar coordinate rasterization;

[0056] Figure 3 It is an operation flow chart for obtaining the ground reference height value of the grid based on the point cloud computing contained in the grid;

[0057] Figure 4 It is an operation flow chart for detecting and obtaining the roadside information of the road to be tested based on the point cloud after filtering out the ground points contained in each grid;

[0058] Figure 5 It is a schematic diagram of the effect of the curb marking points detected by the curb detection method proposed in the embodiment of the present application - a top view;

[0059] Figure 6It is a schematic diagram of the effect of the curb marking points detected by the curb detection method proposed in the embodiment of the present application - a side view;

[0060] Figure 7 is a structural diagram of a curb detection device provided in an embodiment of the present application;

[0061] Figure 8 It is a schematic diagram of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] In the following description, for the purpose of illustration rather than limitation, specific details such as specific system structures, technologies, etc. are proposed, so as to thoroughly understand the embodiments of the present application. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, the detailed description of well-known systems, devices, circuits and methods is omitted to prevent unnecessary details from hindering the description of the present application. In addition, in the description of the present application specification and the attached claims, the terms "first", "second", "third" etc. are only used to distinguish the description, and cannot be interpreted as indicating or suggesting relative importance.

[0063] The embodiment of the present application proposes a road edge detection method, which is suitable for environmental perception during autonomous driving. For example, during the driving of an unmanned vehicle, the road edge detection method can be used to more accurately identify the edge of the road and perceive the drivable area in real time. For more specific technical implementation details of the embodiment of the present application, please refer to the method embodiment described below.

[0064] It should be understood that the execution subjects of the various method embodiments of the present application are various types of terminal devices or servers, such as mobile phones, tablet computers, wearable devices, vehicle-mounted devices, augmented reality (AR) / virtual reality (VR) devices, laptops, ultra-mobile personal computers (UMPC), netbooks, personal digital assistants (PDA), large-screen TVs, etc. The embodiments of the present application do not impose any restrictions on the specific types of the terminal devices and servers.

[0065] See also Figure 1 , shows a roadside detection method provided by an embodiment of the present application, comprising:

[0066] 101. Obtaining point cloud data of the road to be tested;

[0067] First, obtain the point cloud data of the road to be tested. The road to be tested can be any type of road that needs to detect the roadside. The point cloud data can be either a laser radar point cloud or an image point cloud. In actual operation, for the scene where the unmanned vehicle is driving on the road to be tested, more than one laser radar can be installed on the unmanned vehicle to collect the three-dimensional point cloud data of the road in front of the unmanned vehicle through the laser radar.

[0068] 102. Perform rasterization processing on the point cloud data;

[0069] After obtaining the point cloud data of the road to be tested, the point cloud data is then rasterized, that is, the point cloud in the point cloud data is divided into corresponding multiple grids according to the area. The three-dimensional point cloud data scanned by the laser radar is a bunch of data points in the three-dimensional space, and each data point has a corresponding three-dimensional position coordinate. On the one hand, the point cloud data can be rasterized according to the three-dimensional position coordinates, and on the other hand, the point cloud data can be rasterized by polar coordinates.

[0070] The basic principle of polar coordinate rasterization is to convert each point in the point cloud data from three-dimensional coordinates to polar coordinates, and then complete clustering based on the polar coordinates to divide each point into the corresponding sector-shaped grid. Figure 2 As shown, it is a schematic diagram of the grid obtained after the point cloud data is processed by polar coordinate rasterization. Figure 2 The center of is the origin of polar coordinates, and the 24 fan-shaped areas formed by the outward diffusion represent the 24 grids obtained by division. According to the set range and angular resolution, the 3D point cloud can be assigned to the corresponding polar coordinate grid. It should be noted that Figure 2 This is just a simple schematic diagram of polar coordinate grid division. In actual operation, according to the set polar coordinate angle resolution and radial distance resolution, the number of divided grids will be much greater than 24.

[0071] The working characteristics of LiDAR are: the closer the distance, the denser the point cloud, and the farther the distance, the sparser the point cloud. When using polar coordinate grids, the closer the grid is to the origin, the smaller the area, and the farther the grid is from the origin, the larger the area, which is exactly adapted to the working characteristics of LiDAR, so that each grid can contain a certain number of point clouds, thereby improving the effect of point cloud rasterization processing. If the grid is divided into fixed sizes, some grids will contain a large number of point clouds, and some grids will lack point clouds, resulting in unsatisfactory point cloud rasterization processing.

[0072] 103. For each grid obtained after the rasterization process, obtain a ground reference height value of the grid according to the point cloud contained in the grid, and filter out ground points in the point cloud contained in the grid according to the ground reference height value, and the difference between the height of the ground point and the ground reference height value is less than a first threshold;

[0073] After the grid is divided, the following operations are performed for each grid: the ground reference height value corresponding to the grid is calculated based on the point cloud contained in the grid, and the ground points in the point cloud contained in the grid are filtered out based on the ground reference height value. The ground reference height value is used to characterize the true ground height corresponding to a grid. The height of each point in the point cloud contained in the grid can be compared with the ground reference height value. If the difference between the two (the absolute value can be added) is less than the set threshold (the first threshold), the corresponding point can be determined as a ground point in the point cloud. The ground point may cause interference in the subsequent curb detection, so the ground point needs to be filtered out.

[0074] Exemplarily, for grid 1, first calculate the ground reference height value H1 corresponding to the grid based on the point cloud contained in grid 1, and then treat the points in the point cloud contained in grid 1 whose height is close to H1 as ground points for filtering out. For grid 2, first calculate the ground reference height value H2 corresponding to the grid based on the point cloud contained in grid 2, and then treat the points in the point cloud contained in grid 2 whose height is close to H2 as ground points for filtering out. It can be seen that since different grids contain different point clouds, the ground reference height values ​​corresponding to different grids may also be different. In other words, the ground reference height value of a grid is an adaptation value calculated based on the characteristics of the point cloud contained in the grid, which can more accurately represent the true ground height of the area corresponding to the grid. Therefore, this method can more accurately find the ground points in the point cloud contained in each grid for filtering out, thereby improving the accuracy of subsequent curb detection.

[0075] On the other hand, after determining the ground reference height value corresponding to each grid, the obstacle points in the point cloud contained in each grid can also be detected through the ground reference height value corresponding to each grid. Specifically, the points in the point cloud whose height is greater than a certain value of the ground reference height value (for example, 20cm or 10cm) can be determined as obstacle points on the road surface. When outputting the curb detection information, the obstacle point information can be output together so that the unmanned vehicle can perform operations such as obstacle avoidance planning. The obstacle points detected in this process may include some curb points, but these curb points will be identified in the subsequent operations of the embodiments of the present application, and the curb points can also be regarded as special obstacle points, so there is no effect on the curb detection results.

[0076] In one implementation of the embodiment of the present application, Figure 3 As shown, the obtaining of the ground reference height value of the grid according to the point cloud computing contained in the grid may include:

[0077] 301. Selecting a target point cloud corresponding to a partial road surface area of ​​the road to be tested from the point cloud data, wherein the absolute value of the height of each point included in the target point cloud is less than a sixth threshold;

[0078] 302. Calculate an empirical ground height value according to the height of each point included in the target point cloud;

[0079] 303. Calculate the ground height threshold of the grid according to the empirical value of the ground height;

[0080] 304. Extract valid points in the point cloud included in the grid, where the height of the valid points is less than the ground height threshold;

[0081] 305. Perform curve fitting processing on the effective points to obtain a target polynomial equation;

[0082] 306. Solve the target polynomial equation to obtain a ground reference height value of the grid.

[0083] It should be understood that each grid obtained by the above division can obtain the corresponding ground reference height value in the manner of steps 301-306. When solving the ground reference height value of each grid based on the point cloud, it is first necessary to obtain an empirical value of the ground height as a reference value of the ground height. Specifically, the point cloud data corresponding to a partial road surface area of ​​the road to be tested (a passable area, such as a road surface area of ​​a specified size in front of an unmanned vehicle) can be selected and represented by a target point cloud. The absolute value of the height of each point contained in the target point cloud is less than a certain threshold (for example, 30 cm). Next, the empirical value of the ground height can be calculated based on the height of each point contained in the target point cloud. For example, the average value of the height of each point contained in the target point cloud can be calculated as the empirical value of the ground height.

[0084] After obtaining the ground height empirical value, the ground height threshold of the grid can be calculated. On the one hand, the ground height empirical value can be directly used as the ground height threshold of the grid. On the other hand, the ground height empirical value can be adjusted according to a set method. For example, a height adjustment amount can be set, and the adjusted ground height empirical value is used as the ground height threshold of the grid. In an implementation of the embodiment of the present application, after performing polar coordinate rasterization processing on the point cloud data, step 303 may include:

[0085] 3031. Divide the grid into a plurality of grid partitions according to radial distances from the polar coordinate origin;

[0086] 3032. For each of the grid partitions, calculate the height relaxation amount corresponding to the grid partition based on the distribution range of the radial distance between the grid partition and the polar coordinate origin, and determine the sum of the ground height empirical value and the height relaxation amount as the ground height threshold value of the grid partition.

[0087] According to the different radial distances, a certain height relaxation amount can be set. The larger the radial distance, the higher the height relaxation amount is set. For example, the height relaxation amount can be increased by 10cm for every 2m of radial distance. The ground height experience value and the height relaxation amount are added to obtain the corresponding ground height threshold. Exemplarily, for a certain grid, it is first divided into multiple grid partitions according to the radial distance between the grid and the polar coordinate origin. For example, the radial distance is 0-2m for grid partition 1, the radial distance is 2m-4m for grid partition 2, and the radial distance is 4m-6m for grid partition 3... and so on. Then, for grid partition 1, the height relaxation amount is determined to be 0, so the corresponding ground height threshold is equal to the ground height experience value; for grid partition 2, the height relaxation amount is determined to be 10cm, so the corresponding ground height threshold is equal to the ground height experience value + 10cm; for grid partition 3, the height relaxation amount is determined to be 20cm, so the corresponding ground height threshold is equal to the ground height experience value + 20cm, and so on. By setting a stepped ground height threshold, the ground height of an uphill road surface and the ground height of a downhill road surface (the corresponding height relaxation amount is a negative value) can be effectively identified.

[0088] After calculating the ground height threshold of the grid, the points in the point cloud contained in the grid whose height is less than the ground height threshold are regarded as valid points and extracted. This step is equivalent to screening the point cloud in the grid, trying to ensure that the points of the point cloud are near the ground and filtering out non-ground points. It should be noted that the ground height thresholds of different grids can be different, and even for the same grid, the ground height thresholds of different grid partitions can be different. Therefore, when extracting valid points in a certain point cloud, it is necessary to clarify which grid partition of which grid the point cloud is in, so as to determine the corresponding ground height threshold.

[0089] After extracting the valid points in the point cloud contained in the grid, curve fitting processing can be performed on these valid points to obtain a target polynomial equation. Finally, by solving the target polynomial equation, the ground reference height value corresponding to the grid can be obtained. When performing the curve fitting operation, the least squares fitting method or RANSAC curve fitting method can be used. Exemplarily, RANSAC polynomial curve fitting can be used with a fitting coefficient of 3 to fit all valid points in the point cloud contained in the grid to obtain a target polynomial equation. The dependent variable of the target polynomial equation is the height, and the independent variable is the distance from the origin. By solving the target polynomial equation, a height value result can be obtained, and the result is determined as the ground reference height value corresponding to the grid. Among them, the specific method for solving the polynomial equation belongs to the prior art and will not be repeated here.

[0090] The above steps 301-306 are equivalent to constructing a ground height solution model. Compared with directly obtaining the ground reference height value through the scan line feature or grid statistics method, the introduction of the ground height solution model as a priori knowledge takes into account all point clouds in each grid, which can effectively eliminate noise points, thereby obtaining a more accurate ground reference height value. After the ground reference height value of each grid is determined by the method of steps 301-306, the ground points in the point cloud data can be filtered out according to the ground reference height value of each grid.

[0091] 104. Detect and obtain the curb information of the road to be tested according to the point cloud contained in each grid after filtering out the ground points.

[0092] By performing curb detection on the point cloud after filtering out the ground points contained in each grid, that is, the point cloud after filtering out the ground points in the point cloud data, interference from the ground points can be eliminated, thereby improving the accuracy of curb detection.

[0093] In one implementation of the embodiment of the present application, Figure 4 As shown, the detecting and obtaining the roadside information of the road to be tested according to the point cloud after filtering out the ground points contained in each grid may include:

[0094] 401. Perform a clustering operation on the point cloud after filtering out the ground points contained in each of the grids to obtain an initial roadside area;

[0095] 402. Determine a candidate roadside area according to the height difference between each grid contained in the initial roadside area; wherein the height of each grid is equal to the average value of the heights of all points of the point cloud contained therein;

[0096] 403. Select a curb candidate point from the curb candidate area according to the height difference between each point included in the curb candidate area;

[0097] 404. Selecting a roadside marking point from the roadside candidate points according to a preset roadside point determination condition;

[0098] 405. Perform curve fitting processing on the roadside marking points to obtain a roadside marking curve of the road to be tested.

[0099] For step 401, firstly, a clustering operation is performed on the point cloud after filtering out the ground points (corresponding to the point cloud contained in all grids), and the initial curb area can be obtained through the clustering operation. Specifically, density clustering can be adopted. Density clustering starts from the perspective of sample density, checks the connectivity between samples, and continuously expands cluster clusters based on connectable samples to obtain the final clustering result. By performing density clustering on the point cloud, the point cloud can be divided into different clusters, and the area formed by the point cloud of each cluster is an initial curb area. In actual operation, since there are curbs on both sides of the road, the initial curb area obtained here generally includes the left curb area and the right curb area.

[0100] For step 402, after obtaining the initial curb area, the curb candidate area can be further determined based on the height difference between each grid contained therein. The height of each grid can be calculated based on the average value of the heights of all points of the point cloud contained therein. The area where the curb is located has a certain height difference, so the potential area of ​​the curb can be found by detecting the height difference between each grid, and these potential areas are called curb candidate areas.

[0101] Specifically, assuming that the initial roadside area includes M grids, determining the roadside candidate area according to the height difference between the grids included in the initial roadside area may include:

[0102] Traverse every N adjacently arranged grids among the M grids, if the heights of the N grids decrease or increase successively, and the absolute value of the height difference between the first grid and the last grid among the N grids is greater than a second threshold, then determine the N grids as the curb candidate area, 2≤N<M.

[0103] Exemplarily, for the right side of the initial roadside area, the grid is traversed from left to right according to the position of the grid, and for the left side of the initial roadside area, the grid is traversed from right to left according to the position of the grid. The following takes the right side of the roadside area as an example to illustrate the process of determining the candidate roadside area: Assuming that N=3 (the value of N can be reasonably set according to the resolution set when dividing the grid), the grids arranged from left to right in the right side of the roadside area include: grid 1, grid 2, grid 3, grid 4, grid 5, grid 6..., then first select grid 1, grid 2 and grid 3 for detection, if the height of grid 1 is greater than the height of grid 2, the height of grid 2 is greater than the height of grid 3, and the absolute value of the height difference between grid 1 and grid 3 is greater than the set threshold, then grid 1, grid 2 and grid 3 are determined as a candidate roadside area. Next, select grid 2, grid 3 and grid 4 for the same detection until all possible combinations of adjacent grids included in the right side of the roadside area are traversed. At this point, all the candidate roadside regions included in the right roadside region are obtained. The operation performed on the left roadside region is similar, and all the candidate roadside regions included in the left roadside region can also be obtained.

[0104] For step 403, after all the curb candidate areas are obtained, for each curb candidate area, a potential curb point can be selected from the curb candidate area according to the height difference between the points contained therein. The potential curb point can be called a curb candidate point.

[0105] Specifically, assuming that a certain roadside candidate area includes X points, step 403 may include:

[0106] 4031. Sort the X points according to their arrangement positions in the curb candidate area;

[0107] 4032. Select Y points in adjacent order from the X points, 3≤Y<X;

[0108] 4033. If the heights of the Y points decrease or increase successively, and the absolute value of the height difference between the first point and the last point among the Y points is greater than a third threshold, the midpoint of the Y points is determined as the curb candidate point of the curb candidate area.

[0109] First, sort the X points according to their arrangement positions in the curb candidate area. For example, for a curb candidate area included in the right curb area, the points can be sorted from left to right according to their positions. Assuming Y=3, the curb candidate area includes points arranged adjacently from left to right: point 1, point 2, point 3, point 4, point 5, point 6..., then first select point 1, point 2 and point 3 for detection. If the height of point 1 is greater than the height of point 2, the height of point 2 is greater than the height of point 3, and the absolute value of the height difference between point 1 and point 3 is greater than the set threshold, then point 2 is determined as the curb candidate point of the curb candidate area. In the embodiment of the present application, only one curb candidate point is selected for each curb candidate area, so it is not necessary to continue to select point 2, point 3 and point 4 for detection at this time. On the contrary, if the height difference between point 1, point 2 and point 3 does not meet the condition, it can be determined that point 2 is not a candidate point of the roadside candidate area. At this time, it is necessary to continue to select point 2, point 3 and point 4 for detection in the same way until the candidate point of the roadside candidate area is found. For each candidate roadside area, a candidate roadside point can be found. For example, if the right roadside area contains A candidate roadside areas and the left roadside area contains B candidate roadside areas, then A+B candidate roadside points can be detected in step 403.

[0110] For step 404, after obtaining the various roadside candidate points, roadside marking points can be selected from the roadside candidate points according to the set roadside point determination conditions, and the roadside marking points can be regarded as the roadside position points finally detected. Specifically, step 404 may include:

[0111] For each of the curb candidate points, if the difference between the height of the curb candidate point and the ground reference height value of the grid where the curb candidate point is located is greater than a fourth threshold, and the curb candidate point is not within the target detection frame of the point cloud data, then the curb candidate point is determined as a curb marking point.

[0112] Two judgment conditions are set here. The first condition is that the difference between the height of the candidate curb point and the ground reference height value of the grid where the candidate curb point is located is greater than the set threshold. The second condition is that the candidate curb point is not within the target detection frame of the point cloud data (such as a pedestrian detection frame or a vehicle detection frame, which can be obtained using a deep learning-based target detection method). A candidate curb point must meet both conditions at the same time to be determined as the final curb marking point. By setting it in this way, the interference caused by uneven road surface or the presence of other vehicles and pedestrians on the road surface can be removed, greatly improving the accuracy of detecting curb marking points.

[0113] The target detection frame can be generated and integrated into the point cloud data in the following ways:

[0114] (1) acquiring a road image corresponding to the point cloud data;

[0115] (2) performing target detection processing on the road image for non-roadside objects to obtain the target detection frame;

[0116] (3) Fusion of the road image with the target detection frame and the point cloud data.

[0117] In actual operation, a camera can be used to capture images of the road area in front of the unmanned vehicle to obtain a road image corresponding to the point cloud data. Then, a target detection algorithm is used to perform target detection processing on the road image for non-roadside objects (such as pedestrians, vehicles, etc.) to obtain various target detection frames. Finally, the road image with the target detection frame is fused with the point cloud data.

[0118] Figure 5 and Figure 6 : is a schematic diagram of the effect of the roadside marking points detected by the roadside detection method proposed in the embodiment of the present application, wherein Figure 5 It is a top view. Figure 6 is a side view. Figure 5 and Figure 6 In the figure, a part of the original point cloud, the roadside candidate area and the roadside marking points are marked.

[0119] 405. Perform curve fitting processing on the roadside marking points to obtain a roadside marking curve of the road to be tested.

[0120] After obtaining each roadside marking point, curve fitting processing can be performed on these roadside marking points (least square fitting or RANSAC curve fitting can be used) to obtain the roadside marking curve of the road to be tested. Here, there are generally two roadside marking curves obtained: the left roadside curve and the right roadside curve. Specifically, step 405 may include:

[0121] 4051. Perform curve fitting processing on the roadside marking points to obtain the roadside curve of the road to be tested at the current moment;

[0122] 4052. If the difference between the curvature of the curb curve at the current moment and the curvature of the curb curve of the road to be tested at the previous moment is less than a fifth threshold, the curb curve at the current moment is determined as the curb marking curve of the road to be tested and output; wherein, the curb curve of the road to be tested at the previous moment is obtained by performing curb detection on the point cloud data of the road to be tested collected at the previous moment.

[0123] For the scene where the unmanned vehicle is traveling on the road to be tested, the corresponding point cloud data can be collected at each moment, and the roadside curve at that moment can be detected based on the point cloud data. The curve fitted in step 405 is the roadside curve of the road to be tested at the current moment. It is necessary to detect whether the difference between the curvature of the roadside curve at the current moment and the curvature of the roadside curve at the previous moment is less than the set threshold, that is, whether the curvature change amplitude of the roadside curve is within the set range. If the condition is met, it means that the roadside curve at the current moment is qualified and can be output as the final roadside marking curve. If the condition is not met, it means that the roadside curve fitted at the current moment is unqualified (because the interval between two adjacent moments is short, the curvature change amplitude between the roadside curves at two adjacent moments fitted under normal circumstances is small). At this time, the roadside curve before the current moment, the closest to the current moment and meeting the condition can be output as the final roadside marking curve. By setting it in this way, the accuracy of the obtained roadside marking curve can be further improved.

[0124] Exemplarily, assuming that the current time is T5, the time when the unmanned vehicle starts to drive includes T0, T1, T2, T3, T4 and T5. If the difference between the curvature of the curb curve fitted by T5 and the curvature of the curb curve fitted by T4 is less than the set threshold, the curb curve fitted by T5 can be determined as the curb marking curve at time T5 and output. If the difference between the curvature of the curb curve fitted by T5 and the curvature of the curb curve fitted by T4 is greater than or equal to the set threshold, the curb curve closest to time T5 and meeting the conditions is selected and determined as the curb marking curve at time T5 and output. For example, if the curvature change amplitude of the curb curve fitted by T4 and the curb curve fitted by T3 meets the conditions, the curb curve fitted by T4 can be projected into the point cloud data of T5 and output as the curb marking curve of T5.

[0125] In addition, after obtaining each curb marking point, the point cloud height distribution near each curb marking point can also be counted, and the type of curb, such as standard curb, guardrail or fence, can be determined based on the characteristics of the point cloud height distribution. When outputting the curb information, the curb marking curve, the corresponding curb type, and obstacle point information can be output together. Exemplarily, the point cloud height distribution characteristics of different types of curbs are: the point cloud height of the standard curb is concentrated between 10-30cm; the point cloud height of the guardrail is greater than 30cm and is concentrated at a certain height; the point cloud height of the fence is distributed between 10-150cm, and there are point clouds in each height interval, and the point cloud distribution is relatively uniform.

[0126] In an embodiment of the present application, first, the point cloud data of the road to be tested is obtained, and the point cloud data is rasterized to obtain a plurality of grids; then, the ground reference height value corresponding to each grid is calculated based on the point cloud contained in each grid, and the points whose height difference between the point cloud contained in the grid and the corresponding ground reference height value is less than a set threshold are regarded as ground points for filtering out; finally, the curb information of the road to be tested is detected based on the point cloud data after filtering out the ground points. Assuming that a certain area of ​​the road is uneven, the ground reference height value corresponding to the grid where the area is located is an adaptation value obtained based on the point cloud calculation of the uneven road surface, which can more accurately represent the true ground height of the area, and thus can more accurately find the ground points in the point cloud of the area. Afterwards, by filtering out the ground points in the point cloud, the interference of the ground points on the curb detection can be removed, thereby improving the accuracy of the curb detection.

[0127] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0128] A road edge detection method is mainly described above, and a road edge detection device will be described below.

[0129] See also Figure 7 , an embodiment of a road edge detection device in the embodiment of the present application includes:

[0130] Point cloud data acquisition module 701, used to acquire point cloud data of the road to be tested;

[0131] A rasterization processing module 702, used for performing rasterization processing on the point cloud data;

[0132] A ground point filtering module 703 is used to obtain a ground reference height value of each grid obtained after the rasterization process according to the point cloud computing contained in the grid, and filter out ground points in the point cloud contained in the grid according to the ground reference height value, and the difference between the height of the ground point and the ground reference height value is less than a first threshold;

[0133] The curb detection module 704 is used to detect and obtain the curb information of the road to be tested based on the point cloud contained in each grid after filtering out the ground points.

[0134] In one implementation of the embodiment of the present application, the roadside detection module may include:

[0135] A clustering unit, used for performing a clustering operation on the point cloud contained in each grid after filtering out the ground points, to obtain an initial roadside area;

[0136] A curb candidate area determination unit, configured to determine a curb candidate area according to a height difference between each grid contained in the initial curb area; wherein the height of each grid is equal to an average value of the heights of all points of the point cloud contained therein;

[0137] A curb candidate point selection unit, configured to select a curb candidate point from the curb candidate area according to a height difference between each point included in the curb candidate area;

[0138] A roadside marking point selection unit, used for selecting a roadside marking point from the roadside candidate points according to a preset roadside point determination condition;

[0139] The first curve fitting unit is used to perform curve fitting processing on the roadside marking points to obtain a roadside marking curve of the road to be tested.

[0140] Further, the initial roadside area includes M grids, the roadside candidate area includes X points, and the roadside candidate area determination unit may include:

[0141] A grid traversal subunit, used to traverse every N grids arranged adjacently in the M grids, and if the heights of the N grids decrease or increase in sequence, and the absolute value of the height difference between the first grid and the last grid in the N grids is greater than a second threshold, then the N grids are determined as the roadside candidate area, 2≤N<M;

[0142] The roadside candidate point selection unit may include:

[0143] A point sorting subunit, used for sorting the X points according to their arrangement positions in the roadside candidate area;

[0144] A point selection subunit is used to select Y points in adjacent order from the X points, 3≤Y<X;

[0145] The curb candidate point determination subunit is used to determine the midpoint of the Y points as the curb candidate point of the curb candidate area if the heights of the Y points decrease or increase successively and the absolute value of the height difference between the first point and the last point of the Y points is greater than a third threshold.

[0146] Furthermore, the roadside marking point selection unit may include:

[0147] A curb mark point determination subunit is used for, for each of the curb candidate points, if the difference between the height of the curb candidate point and the ground reference height value of the grid where the curb candidate point is located is greater than a fourth threshold, and the curb candidate point is not within the target detection frame of the point cloud data, then determining the curb candidate point as a curb mark point;

[0148] The roadside detection device may further include:

[0149] A road image acquisition module, used to acquire a road image corresponding to the point cloud data;

[0150] A target detection module, used for performing target detection processing of non-roadside objects on the road image to obtain the target detection frame;

[0151] A data fusion module is used to fuse the road image with the target detection frame and the point cloud data.

[0152] Further, the first curve fitting unit may include:

[0153] A curve fitting subunit, used for performing curve fitting processing on the roadside marking points to obtain the roadside curve of the road to be tested at the current moment;

[0154] A curb marking curve determination subunit is used to determine the curb curve at the current moment as the curb marking curve of the road to be tested and output it if the difference between the curvature of the curb curve at the current moment and the curvature of the curb curve of the road to be tested at the previous moment is less than a fifth threshold; wherein the curb curve of the road to be tested at the previous moment is obtained by performing curb detection on the point cloud data of the road to be tested collected at the previous moment.

[0155] In one implementation of the embodiment of the present application, the ground point filtering module may include:

[0156] A target point cloud selection unit, used for selecting a target point cloud corresponding to a partial road surface area of ​​the road to be tested from the point cloud data, wherein the absolute value of the height of each point included in the target point cloud is less than a sixth threshold value;

[0157] A ground height empirical value calculation unit, used to calculate the ground height empirical value according to the height of each point included in the target point cloud;

[0158] A ground height threshold calculation unit, used to calculate the ground height threshold of the grid according to the ground height empirical value;

[0159] A valid point extraction unit, used to extract valid points in the point cloud contained in the grid, wherein the height of the valid points is less than the ground height threshold;

[0160] A second curve fitting unit is used to perform curve fitting processing on the effective points to obtain a target polynomial equation;

[0161] The polynomial equation solving unit is used to solve the target polynomial equation to obtain the ground reference height value of the grid.

[0162] Furthermore, the rasterization processing module may be specifically used to: perform polar coordinate rasterization processing on the point cloud data; the ground height threshold calculation unit may include:

[0163] A grid partitioning subunit, used for dividing the grid into a plurality of grid partitions according to the radial distances from the polar coordinate origin;

[0164] The ground height threshold calculation subunit is used to calculate, for each of the grid partitions, the height relaxation amount corresponding to the grid partition according to the distribution range of the radial distance between the grid partition and the polar coordinate origin, and determine the sum of the ground height empirical value and the height relaxation amount as the ground height threshold of the grid partition.

[0165] The present application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, Figure 1 Any of the curb detection methods represented by .

[0166] The embodiment of the present application also provides a computer program product, when the computer program product is run on a terminal device, the terminal device executes the following Figure 1 Any of the curb detection methods represented by .

[0167] Figure 8 Schematic diagram of a terminal device provided by an embodiment of the present application. Figure 8 As shown, the terminal device 8 of this embodiment includes: a processor 80, a memory 81, and a computer program 82 stored in the memory 81 and executable on the processor 80. When the processor 80 executes the computer program 82, the steps in the above-mentioned embodiments of the motion detection method are implemented, for example Figure 1 Alternatively, when the processor 80 executes the computer program 82, the functions of each module / unit in the above-mentioned device embodiments are realized, for example, Figure 7 Functions of modules 701 to 704 are shown.

[0168] The computer program 82 may be divided into one or more modules / units, which are stored in the memory 81 and executed by the processor 80 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 82 in the terminal device 8.

[0169] The processor 80 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0170] The memory 81 may be an internal storage unit of the terminal device 8, such as a hard disk or memory of the terminal device 8. The memory 81 may also be an external storage device of the terminal device 8, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 8. Further, the memory 81 may also include both an internal storage unit and an external storage device of the terminal device 8. The memory 81 is used to store the computer program and other programs and data required by the terminal device. The memory 81 may also be used to temporarily store data that has been output or is to be output.

[0171] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0172] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0173] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0174] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0175] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, 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 an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0176] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0177] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0178] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0179] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A road edge detection method, characterized in that: include: Obtain point cloud data of the road to be tested; Performing rasterization processing on the point cloud data; For each grid obtained after the rasterization process, a ground reference height value of the grid is obtained according to the point cloud calculation contained in the grid, and ground points in the point cloud contained in the grid are filtered out according to the ground reference height value, and the difference between the height of the ground point and the ground reference height value is less than a first threshold value; Detecting and obtaining the curb information of the road to be tested according to the point cloud contained in each grid after filtering out the ground points; The step of obtaining the ground reference height value of the grid according to the point cloud computing contained in the grid includes: Selecting a target point cloud corresponding to a partial road surface area of ​​the road to be tested from the point cloud data, wherein the absolute value of the height of each point included in the target point cloud is less than a sixth threshold value, and the partial road surface area is a passable area; Calculate the ground height empirical value according to the height of each point included in the target point cloud; Calculate the ground height threshold of the grid according to the empirical value of the ground height; Extracting valid points in the point cloud contained in the grid, wherein the height of the valid points is less than the ground height threshold; Performing curve fitting processing on the effective points to obtain a target polynomial equation; The target polynomial equation is solved to obtain the ground reference height value of the grid.

2. The method according to claim 1, characterized in that The detecting and obtaining the roadside information of the road to be tested according to the point cloud after filtering the ground points contained in each grid comprises: Performing a clustering operation on the point cloud after filtering out the ground points contained in each of the grids to obtain an initial roadside area; Determine a candidate roadside area according to the height difference between each grid contained in the initial roadside area; wherein the height of each grid is equal to the average value of the heights of all points of the point cloud contained therein; Selecting a curb candidate point from the curb candidate area according to the height difference between each point included in the curb candidate area; According to the preset roadside point determination conditions, a roadside marking point is selected from the roadside candidate points; Curve fitting processing is performed on the roadside marking points to obtain a roadside marking curve of the road to be tested.

3. The method according to claim 2, characterized in that The initial roadside area includes M grids, and determining the roadside candidate area according to the height difference between the grids included in the initial roadside area includes: Traversing every N adjacent grids in the M grids, if the heights of the N grids decrease or increase in sequence, and the absolute value of the height difference between the first grid and the last grid in the N grids is greater than a second threshold, then determining the N grids as the roadside candidate area, 2≤N<M; The curb candidate area includes X points, and searching for curb candidate points in the curb candidate area according to the height differences between the points included in the curb candidate area includes: Sorting the X points according to their arrangement positions in the roadside candidate area; Select Y points in adjacent order from the X points, 3≤Y<X; If the heights of the Y points decrease or increase successively, and the absolute value of the height difference between the first point and the last point among the Y points is greater than a third threshold, the midpoint of the Y points is determined as the curb candidate point of the curb candidate area.

4. The method according to claim 2, characterized in that The step of selecting a roadside marking point from the roadside candidate points according to a preset roadside point determination condition comprises: For each of the roadside candidate points, if the difference between the height of the roadside candidate point and the ground reference height value of the grid where the roadside candidate point is located is greater than a fourth threshold, and the roadside candidate point is not within the target detection frame of the point cloud data, then the roadside candidate point is determined as a roadside marking point; The target detection frame is generated and integrated into the point cloud data in the following manner: Acquire a road image corresponding to the point cloud data; Performing target detection processing on the road image for non-roadside objects to obtain the target detection frame; The road image with the target detection frame and the point cloud data are fused.

5. The method according to claim 2, characterized in that The performing curve fitting processing on the roadside marking points to obtain the roadside marking curve of the road to be tested includes: Performing curve fitting processing on the roadside marking points to obtain the roadside curve of the road to be tested at the current moment; If the difference between the curvature of the curb curve at the current moment and the curvature of the curb curve of the road to be tested at the previous moment is less than the fifth threshold, the curb curve at the current moment is determined as the curb marking curve of the road to be tested and output; wherein, the curb curve of the road to be tested at the previous moment is obtained by performing curb detection on the point cloud data of the road to be tested collected at the previous moment.

6. The method according to any one of claims 1 to 5, characterized in that The performing rasterization processing on the point cloud data is specifically: Performing polar coordinate rasterization processing on the point cloud data; The step of calculating the ground height threshold of the grid according to the ground height empirical value includes: Dividing the grid into a plurality of grid partitions according to the radial distances from the polar coordinate origin; For each of the grid partitions, the height relaxation amount corresponding to the grid partition is calculated based on the distribution range of the radial distance between the grid partition and the polar coordinate origin, and the sum of the ground height empirical value and the height relaxation amount is determined as the ground height threshold value of the grid partition.

7. A road edge detection device, characterized in that: include: A point cloud data acquisition module is used to acquire point cloud data of the road to be tested; A rasterization processing module, used for performing rasterization processing on the point cloud data; A ground point filtering module, for obtaining, for each grid obtained after the rasterization process, a ground reference height value of the grid according to the point cloud calculation contained in the grid, and filtering out ground points in the point cloud contained in the grid according to the ground reference height value, wherein the difference between the height of the ground point and the ground reference height value is less than a first threshold value; A curb detection module, used for detecting and obtaining curb information of the road to be tested based on the point cloud contained in each grid after filtering out the ground points; The ground point filtering module comprises: A target point cloud selection unit is used to select a target point cloud corresponding to a partial road surface area of ​​the road to be tested from the point cloud data, wherein the absolute value of the height of each point included in the target point cloud is less than a sixth threshold value, and the partial road surface area is a passable area; A ground height empirical value calculation unit, used to calculate the ground height empirical value according to the height of each point included in the target point cloud; A ground height threshold calculation unit, used to calculate the ground height threshold of the grid according to the ground height empirical value; A valid point extraction unit, used to extract valid points in the point cloud contained in the grid, wherein the height of the valid points is less than the ground height threshold; A second curve fitting unit is used to perform curve fitting processing on the effective points to obtain a target polynomial equation; The polynomial equation solving unit is used to solve the target polynomial equation to obtain the ground reference height value of the grid.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the road edge detection method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the road edge detection method according to any one of claims 1 to 6 is implemented.

Images