A method and apparatus for identifying a ground point cloud

Abstract

The application discloses a method and device for identifying ground point clouds, and relates to the technical field of automatic driving. A specific embodiment of the method comprises: grouping point clouds of multiple radars; for a group in which the point clouds of a blind-filling radar are located: dividing the point clouds of the blind-filling radar into multiple first grids; determining a plane equation of the first grid according to the point clouds in the first grid; identifying whether the point clouds in the first grid are ground point clouds according to the plane equation of the first grid; for a group in which the point clouds of a main radar are located: dividing the point clouds of the main radar into multiple second grids; determining a plane equation of the second grid according to the point clouds in the second grid; dividing a blind area of the main radar into multiple third grids; determining a plane equation of the third grid according to the plane equation of the first grid; and identifying whether the point clouds in the second grid are ground point clouds according to the plane equation of the second grid and the plane equation of the third grid. The embodiment can improve the identification accuracy of ground point clouds.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to a method and device for identifying ground point clouds. BACKGROUND

[0002] An automatic driving vehicle is equipped with multiple sensors to perceive the surrounding environment and objects, and based on the perception results, different strategies such as advancing, waiting, detouring, and avoiding are taken. As an active detection sensor, a laser radar can accurately obtain three-dimensional information of a target, has high resolution, strong anti-interference capability, wide detection range, and other advantages, and has become an indispensable sensor in the automatic driving process.

[0003] Due to the large size of the automatic driving vehicle, no matter where a single laser radar is installed on the vehicle, there will be a detection blind area. Therefore, the automatic driving vehicle often configures a main radar and multiple blind-filling radars to achieve the purpose of information complementation. For example, a main radar is installed on the top of the automatic driving vehicle, and blind-filling radars are installed in the front, rear, left, and right directions respectively.

[0004] In the automatic driving process, it is necessary to identify the point clouds collected by the radars as ground point clouds and non-ground point clouds, and the non-ground point clouds are regarded as potential obstacles for avoidance. The existing point cloud identification method generally only considers identifying the point clouds derived from a single radar, and is not applicable to the case of multiple radars working cooperatively. SUMMARY

[0005] Therefore, the embodiments of the present application provide a method and device for identifying ground point clouds, which initialize the plane equation corresponding to the blind area of the main radar based on the plane equation corresponding to the blind-filling radar in the scene of multiple radars working cooperatively, thereby improving the identification accuracy of the ground point clouds, and having good adaptability to bumpy scenes.

[0006] In a first aspect, the embodiments of the present application provide a method for identifying ground point clouds, comprising:

[0007] grouping the point clouds derived from multiple radars; wherein the point clouds of the main radar and the point clouds of the blind-filling radar are located in different groups, the ground areas corresponding to the point clouds of any two radars in the same group do not coincide, and the point clouds of the blind-filling radar and the point clouds of the main radar are in the same coordinate system;

[0008] for the group in which the point clouds of the blind-filling radar are located: dividing the point clouds of the blind-filling radar into multiple first grids; determining the plane equation of the first grid according to the point clouds in the first grid; and identifying whether the point clouds in the first grid are ground point clouds according to the plane equation of the first grid;

[0009] For the group in which the point cloud of the main radar is located: divide the point cloud of the main radar into a plurality of second grids; determine the plane equation of the second grid according to the point cloud in the second grid; divide the blind area of the main radar into a plurality of third grids; determine the plane equation of the third grid according to the plane equation of the first grid; and identify whether the point cloud in the second grid is ground point cloud according to the plane equation of the second grid and the plane equation of the third grid.

[0010] In a second aspect, an embodiment of the present application provides a device for identifying ground point cloud, comprising:

[0011] The grouping module is configured to group the point clouds from the plurality of radars; wherein the point cloud of the main radar and the point cloud of the blind-filling radar are located in different groups, the ground areas corresponding to the point clouds of any two radars in the same group do not coincide, and the point cloud of the blind-filling radar is in the same coordinate system as the point cloud of the main radar;

[0012] The blind-filling radar identification module is configured to, for the group in which the point cloud of the blind-filling radar is located: divide the point cloud of the blind-filling radar into a plurality of first grids; determine the plane equation of the first grid according to the point cloud in the first grid; and identify whether the point cloud in the first grid is ground point cloud according to the plane equation of the first grid;

[0013] The main radar identification module is configured to, for the group in which the point cloud of the main radar is located: divide the point cloud of the main radar into a plurality of second grids; determine the plane equation of the second grid according to the point cloud in the second grid; divide the blind area of the main radar into a plurality of third grids; determine the plane equation of the third grid according to the plane equation of the first grid; and identify whether the point cloud in the second grid is ground point cloud according to the plane equation of the second grid and the plane equation of the third grid.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, comprising:

[0015] one or more processors;

[0016] a storage device configured to store one or more programs,

[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the method in any of the above embodiments.

[0018] In a fourth aspect, an embodiment of the present application provides a computer readable medium having stored thereon a computer program, wherein the program, when executed by a processor, implements the method in any of the above embodiments.

[0019] An embodiment of the above application has the following advantages or beneficial effects: grouping the point clouds of different radars, in the process of identifying the ground point cloud of the main radar, using the plane equation corresponding to the complementary blind radar to initialize the plane equation corresponding to the blind area of the main radar instead of taking the ground equation as the plane equation corresponding to the blind area of the main radar, which can improve the identification accuracy of the ground point cloud.

[0020] Further effects of the above non-conventional optional mode will be described below in conjunction with the specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings are used to better understand the application and do not constitute undue limitations on the application. Among them:

[0022] Figure 1 is a schematic diagram of the ground point cloud of the main radar and the complementary blind radar in a smooth road surface scene provided by an embodiment of the application;

[0023] Figure 2 is a schematic diagram of the ground point cloud of the main radar and the complementary blind radar in a bumpy road surface scene provided by an embodiment of the application;

[0024] Figure 3 is a flowchart of a method for identifying ground point cloud provided by an embodiment of the application;

[0025] Figure 4 is a flowchart of a method for identifying ground point cloud provided by another embodiment of the application;

[0026] Figure 5 is a schematic diagram of an apparatus for identifying ground point cloud provided by an embodiment of the application;

[0027] Figure 6 is a structural schematic diagram of a computer system of a terminal device or a server suitable for realizing the embodiments of the application. DETAILED DESCRIPTION

[0028] The exemplary embodiments of the application are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the application to help understanding, and should be considered only as exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the application. Also, in order to be clear and concise, the description below omits the description of well-known functions and structures.

[0029] The autonomous vehicle avoids non-ground point clouds as potential obstacles during driving, therefore, accurate identification of ground point clouds and non-ground point clouds is crucial for vehicle driving.

[0030] In scenarios involving multi-radar collaborative operations, the point clouds of the fill-in radar and the main radar can first be transformed to the same coordinate system through calibration relationships, and then the point clouds of different radars can be uniformly identified. During this process, the point clouds of different radars are treated as emitted by the same radar.

[0031] However, vehicles often encounter bumpy road conditions while driving. In such situations, due to the different installation locations of the main radar and the blind spot radar, the trigger times of the sensors often differ. Under bumpy conditions, the radar's observation attitude changes significantly within a short period, easily leading to misalignment of the ground point clouds observed by different radars.

[0032] like Figure 1 The image shows the ground point cloud of the main radar and the blind spot radar in a smooth road surface scenario. On a smooth road surface, based on the calibration relationship, the ground point clouds of the main radar and the blind spot radar can be aligned. At this time, the point clouds of different radars can be processed in a unified manner.

[0033] like Figure 2 The image shows the ground point cloud data of the main radar and the gap-filling radar under bumpy road conditions. On bumpy roads, based on calibration relationships, there will be significant gaps between the ground point clouds of the main radar and the gap-filling radar, as shown below. Figure 2 As shown in the circled area, if the point clouds from different radars are processed uniformly, some ground point clouds within the circled area will be misidentified as non-ground point clouds.

[0034] Therefore, in turbulent scenarios, on the one hand, the identification process of ground point clouds has high requirements for calibration relationships; on the other hand, processing point clouds from different radars in a unified manner based on calibration relationships will result in low accuracy of the identification results.

[0035] In view of this, such as Figure 3 As shown, this embodiment of the invention provides a method for identifying ground point clouds, including:

[0036] Step 301: Group the point clouds from multiple radars.

[0037] Autonomous vehicles are equipped with multiple radars, including several main radars and several blind spot radars. The point clouds collected by different radars are grouped according to the type and location of the radars.

[0038] The point cloud of the blind-filling radar and the point cloud of the main radar are in the same coordinate system. If the point cloud of the main radar and the point cloud of the blind-filling radar are not in the same coordinate system, in order to facilitate subsequent processing, the method further includes: converting the point cloud of the blind-filling radar to the coordinate system of the main radar. Of course, in an actual application scenario, the point cloud of the main radar can also be converted to the coordinate system of the blind-filling radar. It should be noted that the point cloud of the radar can be grouped first, and then the coordinate conversion is performed, or the coordinate conversion can be performed first, and then the point cloud of the radar is grouped.

[0039] The grouping follows the following strategies:

[0040] (1) The point cloud of the main radar and the point cloud of the blind-filling radar are in different groups;

[0041] (2) The ground areas corresponding to the point clouds of any two radars in the same group do not overlap.

[0042] Considering that the point cloud densities of the main radar and the blind-filling radar are different, and the plane equation corresponding to the blind-filling radar is used to initialize the plane equation corresponding to the blind area of the main radar subsequently, the main radar and the blind-filling radar are processed respectively in the embodiment of the application.

[0043] In order to avoid ambiguity in the identification result of the same position, the ground areas corresponding to the point clouds from the two radars do not overlap in the same group. The ground area is determined by the x coordinate and the y coordinate of the point cloud. That is to say, in the same group, the x coordinate and the y coordinate in any point cloud of radar 1 are not completely consistent with the x coordinate and the y coordinate in any point cloud of radar 2.

[0044] The grouping strategy will be described in detail below through three radar layout schemes.

[0045] Layout scheme 1:

[0046] A main radar is installed on the top of the vehicle, and a blind-filling radar is installed in front of, behind, left of and right of the vehicle. In this layout scheme, the horizontal field of view angle of the blind-filling radar is generally less than 180 degrees.

[0047] At this time, the point clouds of the front and rear blind-filling radars can be taken as a first group, the point clouds of the left and right blind-filling radars can be taken as a second group, and the point cloud of the main radar can be taken as a third group.

[0048] Layout scheme 2:

[0049] A main radar is installed on the top of the vehicle, and a blind-filling radar is installed in front of, behind, left of and right of the vehicle. In this layout scheme, the horizontal field of view angle of the blind-filling radar is generally less than 180 degrees.

[0050] At this time, the point cloud of the left front radar can be taken as the first group, the point cloud of the right front radar can be taken as the second group, and the point cloud of the main radar can be taken as the third group.

[0051] Layout scheme 3:

[0052] A main radar is installed on the top of the vehicle, and a blind filling radar is installed in each of the six directions of the front, left front, right front, back, left back and right back of the vehicle. In this layout scheme, the horizontal field of view angle of the blind filling radar is generally less than 180 degrees.

[0053] At this time, the point cloud of the front, left back and right back blind filling radars can be taken as the first group, the point cloud of the left front, right front and back blind filling radars can be taken as the second group, and the point cloud of the main radar can be taken as the third group.

[0054] If there are multiple blind filling radars, the point clouds of the blind filling radars need to be grouped according to the ground areas corresponding to the point clouds of the blind filling radars, so as to ensure that the grouping result meets the above grouping strategy. If there are multiple main radars, the main radars also need to be grouped according to the ground areas corresponding to the point clouds of the main radars.

[0055] In the process of identifying the ground point cloud, the point clouds of the radars in each group can be identified in turn in the order of the first group, the second group, the third group, and so on. If the point clouds of the blind filling radars are grouped into multiple groups, the execution order of each group is not fixed. Taking layout scheme 1 as an example, the point clouds in each group can be identified in turn in the order of the first group, the second group and the third group, or in the order of the second group, the first group and the third group. Alternatively, the first group and the second group can be executed simultaneously.

[0056] Taking layout scheme 1 as an example, in the specific grouping process, the point clouds of the front blind filling radars can be divided into the first group. Since the rear blind filling radars meet strategy (2) with the front blind filling radars, the point clouds of the rear blind filling radars are also divided into the first group. However, the left blind filling radars do not meet strategy (2) with the front blind filling radars, so the left blind filling radars are divided into the second group. Similarly, the right blind filling radars are divided into the second group. The point clouds of the main radar are not in the same group as the point clouds of the blind filling radars, so the point clouds of the main radar are divided into the third group.

[0057] Step 302: For the group in which the point cloud of the blind filling radar is located: divide the point cloud of the blind filling radar into a plurality of first grids; determine the plane equation of the first grid according to the point cloud in the first grid; and identify whether the point cloud in the first grid is a ground point cloud according to the plane equation of the first grid.

[0058] Step 303: for the group in which the point cloud of the main radar is located: dividing the point cloud of the main radar into a plurality of second grids; determining the plane equation of the second grid according to the point cloud in the second grid; dividing the blind area of the main radar into a plurality of third grids; determining the plane equation of the third grid according to the plane equation of the first grid; and identifying whether the point cloud in the second grid is ground point cloud according to the plane equation of the second grid and the plane equation of the third grid.

[0059] In the process of identifying the ground point cloud of the main radar, the plane equation of the blind area of the main radar is initialized by using the plane equation corresponding to the blind-filling radar instead of using the ground equation as the plane equation corresponding to the blind area of the main radar, so that the identification accuracy of the ground point cloud can be improved.

[0060] In an embodiment of the present application, the point cloud of the blind-filling radar is divided into a plurality of first grids in view of the information of the top view angle, including:

[0061] According to the top view angle, the ground area corresponding to the point cloud of the blind-filling radar is divided into a plurality of first grids;

[0062] The first grid to which the point cloud of the blind-filling radar belongs is determined.

[0063] The ground area is determined by the x coordinate and the y coordinate in the point cloud. The first grid in which the point cloud is located is determined according to the x coordinate and the y coordinate of the point cloud.

[0064] Similarly, in an embodiment of the present application, the point cloud of the main radar is divided into a plurality of second grids, including:

[0065] According to the top view angle, the ground area corresponding to the point cloud of the main radar is divided into a plurality of second grids;

[0066] The second grid to which the point cloud of the main radar belongs is determined.

[0067] The point clouds of the main radar and the blind-filling radar are respectively divided into a plurality of grids based on the coordinates of the point clouds, so that more abundant information can be extracted from the point clouds in the unit of the grid, and the accuracy of ground point cloud identification is improved.

[0068] In an embodiment of the present application, the plane equation of the first grid is determined according to the point cloud in the first grid, including:

[0069] If the number of point clouds in the first grid is not less than 3, the plane equation of the first grid is determined based on the random sample consensus algorithm.

[0070] In addition to the RANSAC (Random Sample Consensus) algorithm, the SVD (Singular Value Decomposition) algorithm can also be used to determine the plane equation of the first grid.

[0071] The plane equation of the first grid can be efficiently determined by RANSAC, thereby improving the identification efficiency.

[0072] If the number of point clouds in the first grid is less than 3, the plane equation of the first grid is as shown in Equation 1.

[0073] z+w′=0 (1)

[0074] wherein z is used to represent the coordinate value of the point cloud in the z-axis direction, and w′ is used to represent the vertical distance of the main radar from the ground.

[0075] The plane equation of the first grid is as shown in Equation 2.

[0076] ax+by+cz+w=0 (2)

[0077] wherein (x, y, z) is the coordinate of the point cloud, (a, b, c) is the plane normal vector of the first grid, and w is used to represent the offset.

[0078] The plane of the first grid refers to the plane formed by the point cloud located in the first grid.

[0079] Similarly, according to the point cloud in the second grid, the plane equation of the second grid is determined, including:

[0080] If the number of point clouds in the second grid is not less than 3, the plane equation of the second grid is determined based on the random sample consensus algorithm.

[0081] The determination process of the plane equation of the second grid is the same as that of the plane equation of the first grid. The SVD algorithm can also be used, and Equation 1 and Equation 2 are also applicable to the plane equation of the second grid, which will not be described here.

[0082] In an embodiment of the present application, according to the plane equation of the first grid, the plane equation of the third grid is determined, including:

[0083] The plane equation of the third grid is determined as the plane equation of the first grid closest to the third grid.

[0084] Considering the similarity of adjacent ground points, the embodiment of the present application initializes the third grid in the main radar blind area based on the plane equation of the first grid instead of using the theoretical ground equation as shown in formula 1 as the plane equation of the third grid, and in the process of identifying the point cloud of the main radar, the information of the blind-filling radar is fully utilized, so that the identification accuracy of the ground point cloud can be improved.

[0085] It should be noted that using the plane equation of the first grid closest to the third grid is only a preferred manner, and the plane equation of the third grid can also be determined based on the plane equations of other adjacent first grids, and the plane equation of the third grid can also be calculated from the plane equations of multiple first grids.

[0086] In an embodiment of the present application, the size of the first grid is smaller than the size of the second grid, and the size of the second grid is equal to the size of the third grid.

[0087] In actual application scenarios, the detection distance of the blind-filling radar is small, but the density of its point cloud is high; the detection distance of the main radar is large, but the density of its point cloud is low.

[0088] Therefore, the size of the second grid is generally larger than the size of the first grid. Since the second grid and the third grid are used in the process of identifying the point cloud of the main radar, the size of the third grid is generally consistent with the size of the second grid. For example, the effective range of the blind-filling radar is within 10 meters, the size of the first grid is 0.2m x 0.2m, the effective range of the main radar is 8-80 meters, and the blind area is within 8 meters, and the size of the second grid and the third grid is 0.5m x 0.2m. Of course, the third grid can also have the same size as the first grid.

[0089] In an embodiment of the present application, according to the plane equation of the first grid, whether the point cloud in the first grid is a ground point cloud is identified, comprising:

[0090] According to the plane equation of the first grid, whether the first grid is a ground grid is identified;

[0091] According to the identification result and the coordinates of the point cloud in the first grid, whether the point cloud in the first grid is a ground point cloud is identified.

[0092] In actual application scenarios, the point cloud in the ground grid can be directly regarded as a ground point cloud. However, in order to further improve the identification accuracy, the embodiment of the present application further identifies each point cloud in the first grid based on the coordinates of the point cloud in the first grid. Specifically, whether the point cloud in the first grid is a ground point cloud can be identified based on the coordinates of the point cloud in the first grid and the plane equation of the first grid belonging to the ground grid. The specific implementation manner will be described in subsequent embodiments.

[0093] In an embodiment of the present application, according to the plane equation of the first grid, whether the first grid is a ground grid is identified, comprising:

[0094] According to the plane equation of the first grid, whether the two adjacent first grids are similar is determined;

[0095] According to the determination result and the position of the autonomous vehicle, whether the first grid is a ground grid is identified;

[0096] Wherein, the ground area where the autonomous vehicle is located is a target grid, and the target grid belongs to the ground grid; if the two adjacent first grids are similar, and one of the first grids belongs to the ground grid, the other first grid also belongs to the ground grid.

[0097] The embodiment of the present application divides the first grids into two categories by determining whether the first grids are similar, and determines which category is the ground grid and which category is the non-ground grid based on the ground area where the autonomous vehicle is located. The plane equation of the target grid can be a theoretical ground equation, as shown in formula 1. If a first grid is similar to the target grid, the first grid is a ground grid, otherwise, it is a non-ground grid. Through the embodiment of the present application, whether each first grid belongs to the ground grid can be accurately identified.

[0098] In an embodiment of the present application, according to the plane equation of the first grid, whether the two adjacent first grids are similar is determined, comprising:

[0099] If the included angle of the plane normal vectors of the two adjacent first grids is less than a preset angle threshold, and the difference of the offsets is less than a preset offset threshold, it is determined that the two adjacent first grids are similar.

[0100] The embodiment of the present application considers the included angle of the plane normal vector and the difference of the offset when measuring whether they are similar, which can further improve the identification accuracy. In actual application scenarios, whether two first grids are similar can be determined based on only the included angle of the plane normal vector or the difference of the offset. The angle threshold and the offset threshold can be preset according to the road surface condition. Compared with a flat road surface, the angle threshold and the offset threshold of a bumpy road surface are slightly larger. For example, the angle threshold is 10 degrees, and the offset threshold is 0.4.

[0101] In an embodiment of the present application, according to the identification result and the coordinates of the point cloud in the first grid, whether the point cloud in the first grid is a ground point cloud is identified, comprising:

[0102] If the first grid is a ground grid, for the current point cloud in the first grid: according to the coordinates of the current point cloud and the plane equation of the first grid, the distance between the current point and the plane of the first grid is calculated, and if the distance between the current point and the plane of the first grid is less than a preset distance threshold, the current point cloud is a ground point cloud;

[0103] If the first grid is a non-ground grid, a target ground grid closest to the first grid is determined in the current group; for the current point cloud in the first grid: according to the coordinates of the current point cloud and the plane equation of the target ground grid, the distance between the current point and the plane of the target ground grid is calculated, and if the distance between the current point and the plane of the target ground grid is less than the distance threshold, the current point cloud is a ground point cloud.

[0104] The distance threshold can be pre-set according to the road surface condition, for example, the distance threshold is 0.08 meters. The embodiment of the application determines whether the point cloud is a ground point cloud based on the distance between the point cloud and the plane of the ground grid. In actual application scenarios, if the first grid is a ground grid, the current point cloud can also be identified according to other ground grids closest to the first grid in the current group. If the first grid is a non-ground grid, the current point cloud can also be identified according to multiple other ground grids in the current group. Therefore, there are various implementation manners for identifying the current point cloud, which will not be listed one by one here.

[0105] In an embodiment of the application, whether the point cloud in the second grid is a ground point cloud is identified according to the plane equation of the second grid and the plane equation of the third grid, comprising:

[0106] According to the plane equation of the second grid and the plane equation of the third grid, whether the second grid is a ground grid is identified;

[0107] According to the identification result and the coordinates of the point cloud in the second grid, whether the point cloud in the second grid is a ground point cloud is identified.

[0108] Similar to the identification process of the first grid, in actual application scenarios, the point cloud in the ground grid can be directly regarded as a ground point cloud. However, in order to further improve the identification accuracy, the embodiment of the application further identifies each point cloud in the second grid based on the coordinates of the point cloud in the second grid. Specifically, whether the point cloud in the second grid is a ground point cloud can be identified based on the coordinates of the point cloud in the second grid and the plane equation of the second grid belonging to the ground grid. The specific implementation manner will be described in subsequent embodiments.

[0109] In an embodiment of the application, whether the second grid is a ground grid is identified according to the plane equation of the second grid and the plane equation of the third grid, comprising:

[0110] According to the plane equation of the second grid and the plane equation of the third grid, whether the two adjacent main radar grids are similar is judged; the main radar grids comprise: the second grid and the third grid;

[0111] According to the judgment result and the position of the autonomous vehicle, whether the second grid is a ground grid is identified;

[0112] The ground area where the autonomous vehicle is located is a target grid, and the target grid belongs to a ground grid; if two adjacent main radar grids are similar and one of the main radar grids belongs to a ground grid, the other main radar grid also belongs to a ground grid.

[0113] The main radar grid includes two types, a second grid located in the main radar range and a third grid located in the main radar blind area. The two adjacent main radar grids can be two second grids, can be two third grids, or can be one second grid and one third grid.

[0114] Similar to the first grid, the embodiment of the application divides the main radar grid into two categories by judging whether the main radar grids are similar, and determines which category is a ground grid and which category is a non-ground grid based on the ground area where the autonomous vehicle is located. The plane equation of the target grid can be a theoretical ground equation, as shown in formula 1. If a main radar grid is similar to the target grid, the main radar grid is a ground grid, otherwise, it is a non-ground grid. Through the embodiment of the application, whether each second grid belongs to a ground grid can be accurately identified.

[0115] In an embodiment of the application, whether the two adjacent main radar grids are similar is judged according to the plane equation of the second grid and the plane equation of the third grid, including:

[0116] If the included angle of the plane normal vectors of the two adjacent main radar grids is less than a preset angle threshold, and the difference between the offsets is less than a preset offset threshold, it is determined that the two adjacent main radar grids are similar.

[0117] Similar to the first grid, the embodiment of the application considers the included angle of the plane normal vector and the difference between the offsets when measuring similarity, which can further improve the recognition accuracy. In actual application scenarios, whether two main radar grids are similar can be determined based on only the included angle of the plane normal vector or the difference between the offsets. The angle threshold and the offset threshold can be preset according to the road surface condition. Compared with a flat road, the angle threshold and the offset threshold of a bumpy road are slightly larger. For example, the angle threshold is 10 degrees and the offset threshold is 0.4.

[0118] In an embodiment of the application, whether the point cloud in the second grid is a ground point cloud is identified according to the recognition result and the coordinates of the point cloud in the second grid, including:

[0119] If the second grid is a ground grid, for the current point cloud in the second grid: according to the coordinates of the current point cloud and the plane equation of the second grid, the distance between the current point and the plane of the second grid is calculated, and if the distance between the current point and the plane of the second grid is less than a preset distance threshold, the current point cloud is a ground point cloud.

[0120] If the second grid is a non-ground grid, a target ground grid closest to the second grid is determined in the current group; for the current point cloud in the second grid: according to the coordinates of the current point cloud and the plane equation of the target ground grid, the distance between the current point and the plane of the target ground grid is calculated, and if the distance between the current point and the plane of the target ground grid is less than the distance threshold, the current point cloud is a ground point cloud.

[0121] Similar to the first grid, the distance threshold can be pre-set according to the road surface condition, for example, the distance threshold is 0.08 meters. The embodiment of the application determines whether the point cloud is a ground point cloud based on the distance between the point cloud and the plane of the ground grid. In actual application scenarios, if the second grid is a ground grid, the current point cloud can also be identified according to other ground grids closest to the second grid in the current group. If the second grid is a non-ground grid, the current point cloud can also be identified according to multiple other ground grids in the current group. Therefore, there are various implementation manners for identifying the current point cloud, which will not be listed one by one here.

[0122] As shown in FIG. 1, Figure 4 the embodiment of the application takes layout scheme 1 as an example to explain the method for identifying the ground point cloud in detail, which comprises:

[0123] Step 401: The point clouds from one main radar and four blind-filling radars are divided into three groups, wherein the point clouds of the front and rear blind-filling radars are taken as the first group, the point clouds of the left and right blind-filling radars are taken as the second group, and the point clouds of the main radar are taken as the third group.

[0124] Step 402: for the first group of point clouds: the point clouds of the front and rear blind-filling radars are divided into multiple first grids; the plane equation of the first grid is determined according to the point clouds in the first grid; and whether the point clouds in the first grid are ground point clouds is identified according to the plane equation of the first grid.

[0125] Step 403: for the second group of point clouds: the point clouds of the left and right blind-filling radars are divided into multiple first grids; the plane equation of the first grid is determined according to the point clouds in the first grid; and whether the point clouds in the first grid are ground point clouds is identified according to the plane equation of the first grid.

[0126] Step 404: for the third group of point clouds: the point clouds of the main radar are divided into multiple second grids; the plane equation of the second grid is determined according to the point clouds in the second grid; the blind area of the main radar is divided into multiple third grids; the plane equation of the third grid is determined according to the plane equation of the first grid; and whether the point clouds in the second grid are ground point clouds is identified according to the plane equation of the second grid and the plane equation of the third grid.

[0127] The embodiment of the present application respectively identifies the point clouds of the main radar and the blind-filling radar, reduces the dependence on the calibration relationship, determines the plane equation corresponding to the blind area of the main radar based on the plane equation corresponding to the blind-filling radar, and improves the accuracy of point cloud identification. In particular, in the bumpy scene, for the intersection area of the main radar and the blind-filling radar, the embodiment of the present application has good adaptability and can effectively reduce the error rate.

[0128] As shown in Figure 5 The embodiment of the present application provides a device for identifying ground point clouds, which comprises:

[0129] The grouping module 501 is configured to group the point clouds from multiple radars; wherein the point clouds of the main radar and the point clouds of the blind-filling radar are located in different groups, the ground areas corresponding to any two radar point clouds in the same group do not coincide, and the point clouds of the blind-filling radar and the point clouds of the main radar are in the same coordinate system.

[0130] The blind-filling radar identification module 502 is configured to, for the group in which the point clouds of the blind-filling radar are located: divide the point clouds of the blind-filling radar into multiple first grids; determine the plane equation of the first grid according to the point clouds in the first grid; and identify whether the point clouds in the first grid are ground point clouds according to the plane equation of the first grid.

[0131] The main radar identification module 503 is configured to, for the group in which the point clouds of the main radar are located: divide the point clouds of the main radar into multiple second grids; determine the plane equation of the second grid according to the point clouds in the second grid; divide the blind area of the main radar into multiple third grids; determine the plane equation of the third grid according to the plane equation of the first grid; and identify whether the point clouds in the second grid are ground point clouds according to the plane equation of the second grid and the plane equation of the third grid.

[0132] In an embodiment of the present application, the blind-filling radar identification module 502 is configured to divide the ground area corresponding to the point clouds of the blind-filling radar into multiple first grids according to the overhead perspective; and determine the first grid to which the point clouds of the blind-filling radar belong.

[0133] In an embodiment of the present application, the main radar identification module 503 is configured to divide the ground area corresponding to the point clouds of the main radar into multiple second grids according to the overhead perspective; and determine the second grid to which the point clouds of the main radar belong.

[0134] In an embodiment of the present application, the blind-filling radar identification module 502 is configured to, if the number of point clouds in the first grid is not less than 3, determine the plane equation of the first grid based on the random sample consensus algorithm.

[0135] In an embodiment of the present application, the main radar identification module 503 is configured to determine the plane equation of the second grid based on a random sample consensus algorithm if the number of point clouds in the second grid is not less than 3.

[0136] In an embodiment of the present application, the main radar identification module 503 is configured to determine the plane equation of the third grid as the plane equation of the first grid closest to the third grid.

[0137] In an embodiment of the present application, the size of the first grid is smaller than the size of the second grid, and the size of the second grid is equal to the size of the third grid.

[0138] In an embodiment of the present application, the blind-filling radar identification module 502 is configured to identify whether the first grid is a ground grid according to the plane equation of the first grid, and identify whether the point cloud in the first grid is a ground point cloud according to the identification result and the coordinates of the point cloud in the first grid.

[0139] In an embodiment of the present application, the blind-filling radar identification module 502 is configured to judge whether the two adjacent first grids are similar according to the plane equation of the first grid, and identify whether the first grid is a ground grid according to the judgment result and the position of the autonomous vehicle; wherein the ground area where the autonomous vehicle is located is a target grid, and the target grid belongs to the ground grid; if the two adjacent first grids are similar and one of the first grids belongs to the ground grid, the other first grid also belongs to the ground grid.

[0140] In an embodiment of the present application, the blind-filling radar identification module 502 is configured to determine that the two adjacent first grids are similar if the included angle of the normal vectors of the two adjacent first grids is less than a preset angle threshold and the difference of the offsets is less than a preset offset threshold.

[0141] In an embodiment of the present application, the blind-filling radar identification module 502 is configured to, if the first grid is a ground grid, for the current point cloud in the first grid: calculate the distance between the current point and the plane of the first grid according to the coordinates of the current point cloud and the plane equation of the first grid, and if the distance between the current point and the plane of the first grid is less than a preset distance threshold, the current point cloud is a ground point cloud.

[0142] If the first grid is a non-ground grid, determine the target ground grid closest to the first grid in the current group; for the current point cloud in the first grid: calculate the distance between the current point and the plane of the target ground grid according to the coordinates of the current point cloud and the plane equation of the target ground grid, and if the distance between the current point and the plane of the target ground grid is less than the distance threshold, the current point cloud is a ground point cloud.

[0143] In an embodiment of the present application, the main radar identification module 503 is configured to identify whether the second grid is a ground grid according to the plane equation of the second grid and the plane equation of the third grid; and identify whether the point cloud in the second grid is a ground point cloud according to the identification result and the coordinates of the point cloud in the second grid.

[0144] In an embodiment of the present application, the main radar identification module 503 is configured to determine whether the two adjacent main radar grids are similar according to the plane equation of the second grid and the plane equation of the third grid; the main radar grids include the second grid and the third grid; identify whether the second grid is a ground grid according to the determination result and the position of the autonomous vehicle; wherein the ground area where the autonomous vehicle is located is a target grid, and the target grid belongs to the ground grid; if the two adjacent main radar grids are similar and one of the main radar grids belongs to the ground grid, the other main radar grid also belongs to the ground grid.

[0145] In an embodiment of the present application, the main radar identification module 503 is configured to determine that the two adjacent main radar grids are similar if the included angle of the plane normal vectors of the two adjacent main radar grids is less than a preset angle threshold and the difference of the offset amounts is less than a preset offset threshold.

[0146] In an embodiment of the present application, the main radar identification module 503 is configured to, if the second grid is a ground grid, for the current point cloud in the second grid: calculate the distance between the current point and the plane of the second grid according to the coordinates of the current point cloud and the plane equation of the second grid, and if the distance between the current point and the plane of the second grid is less than a preset distance threshold, the current point cloud is a ground point cloud.

[0147] If the second grid is a non-ground grid, determine the target ground grid closest to the second grid in the current group; for the current point cloud in the second grid: calculate the distance between the current point and the plane of the target ground grid according to the coordinates of the current point cloud and the plane equation of the target ground grid, and if the distance between the current point and the plane of the target ground grid is less than a distance threshold, the current point cloud is a ground point cloud.

[0148] An electronic device is provided in an embodiment of the present application, which includes:

[0149] one or more processors;

[0150] a storage device configured to store one or more programs,

[0151] When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any of the above embodiments.

[0152] This invention provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the above embodiments.

[0153] The following is for reference. Figure 6 It shows a schematic diagram of the structure of a computer system 600 suitable for implementing a terminal device of the present invention. Figure 6 The terminal device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0154] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 602 or programs loaded from storage section 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the system 600. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0155] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.

[0156] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined above in the system of this invention.

[0157] It should be noted that the computer-readable medium shown in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or component. In the present application, the computer-readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to an electromagnetic signal, an optical signal or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, which can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or component. The program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0158] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment, or a portion of code that contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different orders than that shown in the figures. For example, two blocks that are shown in succession can actually be executed substantially concurrently, or they can sometimes be executed in reverse order, depending on the functionality involved. It should also be noted that each block in the block diagrams or flowcharts, and combinations of blocks in the block diagrams or flowcharts, can be implemented by a dedicated hardware-based system that performs specified functions or operations, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0159] The modules described in the embodiments of the present application can be implemented in the form of software or in the form of hardware. The described modules can also be arranged in a processor, for example, a processor can be described as including a sending module, an obtaining module, a determining module and a first processing module. In some cases, the names of these modules do not constitute a limitation on the modules themselves, for example, the sending module can also be described as "a module that sends a picture obtaining request to a connected server".

[0160] The specific embodiments described above do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can occur depending on design requirements and other factors. Any modification, equivalent replacement and improvement made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method of identifying a ground point cloud, characterized in that, The method comprises: grouping point clouds from a plurality of radars, the radars comprising a main radar and a blind-filling radar; wherein the point cloud of the main radar and the point cloud of the blind-filling radar are in different groups, the ground areas corresponding to the point clouds of any two radars in the same group do not overlap, and the point cloud of the blind-filling radar is in the same coordinate system as the point cloud of the main radar; for the group in which the point cloud of the blind-filling radar is located, dividing the point cloud of the blind-filling radar into a plurality of first grids; determining the plane equation of the first grid according to the point cloud in the first grid; and identifying whether the point cloud in the first grid is ground point cloud according to the plane equation of the first grid; for the group in which the point cloud of the main radar is located, dividing the point cloud of the main radar into a plurality of second grids; determining the plane equation of the second grid according to the point cloud in the second grid; dividing the blind area of the main radar into a plurality of third grids; and determining the plane equation of the third grid according to the plane equation of the first grid; identifying whether the point cloud in the second grid is ground point cloud according to the plane equation of the second grid and the plane equation of the third grid.

2. The method of claim 1, wherein: dividing the point cloud of the blind-filling radar into a plurality of first grids comprises: dividing the ground area corresponding to the point cloud of the blind-filling radar into a plurality of first grids according to a top-down perspective; determining the first grid to which the point cloud of the blind-filling radar belongs; and / or dividing the point cloud of the main radar into a plurality of second grids comprises: dividing the ground area corresponding to the point cloud of the main radar into a plurality of second grids according to a top-down perspective; determining the second grid to which the point cloud of the main radar belongs.

3. The method of claim 1, wherein: determining the plane equation of the first grid according to the point cloud in the first grid comprises: if the number of point clouds in the first grid is not less than 3, determining the plane equation of the first grid based on a random sample consensus algorithm; and / or determining the plane equation of the second grid according to the point cloud in the second grid comprises: if the number of point clouds in the second grid is not less than 3, determining the plane equation of the second grid based on a random sample consensus algorithm.

4. The method of claim 1, wherein: determining the plane equation of the third grid according to the plane equation of the first grid comprises: determining the plane equation of the third grid as the plane equation of the first grid closest to the third grid.

5. The method of claim 1, wherein: the size of the first grid is smaller than the size of the second grid, and the size of the second grid is equal to the size of the third grid.

6. The method of claim 1, wherein: identifying whether the point cloud in the first grid is ground point cloud according to the plane equation of the first grid comprises: identifying whether the first grid is a ground grid according to the plane equation of the first grid; identifying whether the point cloud in the first grid is ground point cloud according to the identification result and the coordinates of the point cloud in the first grid. ​ ​ 7. The method of claim 6, wherein the identifying whether the first grid is a ground grid based on the plane equation of the first grid comprises: determining whether two adjacent first grids are similar based on the plane equation of the first grid; and identifying whether the first grid is a ground grid based on the determination result and a position of the autonomous vehicle.

8. The method of claim 7, wherein the determining whether two adjacent first grids are similar based on the plane equation of the first grid comprises: determining whether an included angle between normal vectors of two adjacent first grids is less than a preset angle threshold and a difference between offset values is less than a preset offset threshold.

9. The method of claim 6, wherein the identifying whether the point cloud in the first grid is a ground point cloud based on the identification result and the coordinates of the point cloud in the first grid comprises: if the first grid is a ground grid, determining, for a current point cloud in the first grid, whether the current point cloud is a ground point cloud based on the coordinates of the current point cloud and the plane equation of the first grid; and if the first grid is a non-ground grid, determining a target ground grid closest to the first grid in a current group, determining, for a current point cloud in the first grid, whether the current point cloud is a ground point cloud based on the coordinates of the current point cloud and the plane equation of the target ground grid.

10. The method of claim 1, wherein the identifying whether the point cloud in the second grid is a ground point cloud based on the plane equation of the second grid and the plane equation of the third grid comprises: identifying whether the second grid is a ground grid based on the plane equation of the second grid and the plane equation of the third grid; and identifying whether the point cloud in the second grid is a ground point cloud based on the identification result and the coordinates of the point cloud in the second grid.

11. The method of claim 10, wherein the identifying whether the second grid is a ground grid based on the plane equation of the second grid and the plane equation of the third grid comprises: determining whether two adjacent main radar grids are similar based on the plane equation of the second grid and the plane equation of the third grid; and identifying whether the second grid is a ground grid based on the determination result and a position of the autonomous vehicle. ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ The ground area where the autonomous vehicle is located is a target grid, and the target grid belongs to a ground grid; if two adjacent main radar grids are similar, and one of the main radar grids belongs to a ground grid, then the other main radar grid also belongs to a ground grid.

12. The method of claim 11, wherein, determining whether two adjacent main radar grids are similar according to the plane equation of the second grid and the plane equation of the third grid comprises: if the included angle of the plane normal vectors of the two adjacent main radar grids is less than a preset angle threshold, and the difference between the offset amounts is less than a preset offset threshold, then it is determined that the two adjacent main radar grids are similar.

13. The method of claim 10, wherein, identifying whether the point cloud in the second grid is a ground point cloud according to the identification result and the coordinates of the point cloud in the second grid comprises: if the second grid is a ground grid, then for a current point cloud in the second grid: according to the coordinates of the current point cloud and the plane equation of the second grid, the distance between the current point and the plane of the second grid is calculated, and if the distance between the current point and the plane of the second grid is less than a preset distance threshold, then the current point cloud is a ground point cloud; if the second grid is a non-ground grid, then the target ground grid closest to the second grid is determined in the current group; for a current point cloud in the second grid: according to the coordinates of the current point cloud and the plane equation of the target ground grid, the distance between the current point and the plane of the target ground grid is calculated, and if the distance between the current point and the plane of the target ground grid is less than the distance threshold, then the current point cloud is a ground point cloud.

14. An apparatus for identifying a ground point cloud, the apparatus comprising: comprises: a grouping module configured to group point clouds derived from a plurality of radars, the radars including main radars and blind-filling radars; wherein the point clouds of the main radars and the point clouds of the blind-filling radars are located in different groups, the ground areas corresponding to the point clouds of any two radars in the same group do not overlap, and the point clouds of the blind-filling radars and the point clouds of the main radars are in the same coordinate system; a blind-filling radar identification module configured to, for the group in which the point clouds of the blind-filling radars are located, divide the point clouds of the blind-filling radars into a plurality of first grids; determine a plane equation of the first grid according to the point clouds in the first grid; and identify whether the point clouds in the first grid are ground point clouds according to the plane equation of the first grid; a main radar identification module configured to, for the group in which the point clouds of the main radars are located, divide the point clouds of the main radars into a plurality of second grids; determine a plane equation of the second grid according to the point clouds in the second grid; divide the blind area of the main radars into a plurality of third grids; determine a plane equation of the third grid according to the plane equation of the first grid; and identify whether the point clouds in the second grid are ground point clouds according to the plane equation of the second grid and the plane equation of the third grid.

15. An electronic device, comprising: comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement a method as claimed in any of claims 1-13.

16. A computer readable medium having stored thereon a computer program, characterized in that, The program, which when executed by a processor, implements a method as claimed in any of claims 1-13.

Images