Road data analysis method and device, electronic equipment, storage medium and vehicle

Abstract

The invention relates to a road data analysis method and device, electronic equipment, a storage medium and a vehicle, and the method comprises the steps: carrying out the clustering of road point clouds in a BEV road perception image, obtaining N point clusters, taking one point cluster as a target point cluster, carrying out the singular value decomposition of the target point cluster, obtaining a first singular value and a second singular value, and carrying out the analysis of the first singular value and the second singular value; determining a road type indicated by the target point cluster based on the difference between the first singular value and the second singular value, determining a target road line indicated by the target point cluster as a first target road line, and determining a first interior point rate of a point cloud corresponding to the first target road line based on a random sampling consistency algorithm corresponding to the road type; and under the condition that the first interior point rate is smaller than or equal to an interior point rate threshold value corresponding to the road type, determining that the first target road line is a deformed line. The method can accurately analyze the abnormal line in the road data.

Description

Technical Field

[0001] The present application relates to the field of automobile technology, and in particular to a road data analysis method, device, electronic equipment, storage medium and vehicle. Background Art

[0002] In the field of autonomous driving, autonomous vehicles determine the vehicle's driving path based on high-precision maps. Therefore, high-precision maps are an important technology for achieving safe autonomous driving. In order to improve the accuracy of map production, when obtaining road data for map production, it is necessary to identify and process abnormal lines in the road data, such as identifying and removing deformed lines in the road lines (lane lines and road edge lines) indicated in the road data, or identifying and removing other outlier lines.

[0003] At present, before eliminating abnormal lines in road lines indicated by road data, it is necessary to first analyze which ones in the road data are abnormal lines. The current solution for analyzing whether road lines in road data are abnormal lines is: for each road line indicated by the road data, first calculate the curvature of every three points in the road line, and then accumulate and calculate the average curvature of the road line. When the average curvature is greater than a threshold (indicating that the degree of distortion of the road line is large), determine that the road line is an abnormal line.

[0004] However, the above-mentioned solution of analyzing whether a road line in road data is an abnormal line is likely to identify a normal curve line as an abnormal line, and the accuracy of abnormal line identification is low, resulting in a large deviation between the map produced based on the road data and the actual road information, and cannot be used for vehicle control and display. Therefore, how to more accurately analyze whether there is an abnormal line in the road data has become an urgent problem to be solved. Summary of the invention

[0005] The present application provides a road data analysis method, device, electronic device, storage medium and vehicle, which can improve the accuracy of abnormal line removal to improve the accuracy of maps produced based on road perception images.

[0006] In a first aspect, an embodiment of the present application provides a road data analysis method, the method comprising:

[0007] Clustering the road point cloud in the BEV road perception image to obtain N point clusters, where N is a positive integer, each point cluster in the N point clusters is used to indicate a different road segment, and the point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, where the target road line is a lane line or a road edge line;

[0008] One of the N point clusters is used as the target point cluster, and the target point cluster is subjected to singular value decomposition to obtain a first singular value and a second singular value, wherein the first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster;

[0009] Determining a road type of a target road segment indicated by the target point cluster based on a difference between the first singular value and the second singular value;

[0010] Determine a target road line indicated by the target point cluster as a first target road line, and determine a first inlier rate of a point cloud corresponding to the first target road line based on an inlier rate algorithm corresponding to the road type;

[0011] When the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line.

[0012] In a second aspect, an embodiment of the present application provides a road data analysis device, the device comprising:

[0013] A clustering unit, used for clustering the road point cloud in the BEV road perception image according to the point spacing to obtain N point clusters, where N is a positive integer, each point cluster in the N point clusters is used to indicate a different road segment, and the point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, where the target road line is a lane line or a road edge line;

[0014] A singular value decomposition unit is used to take one of the N point clusters as a target point cluster, perform singular value decomposition on the target point cluster, and obtain a first singular value and a second singular value, wherein the first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster;

[0015] a determining unit, configured to determine a road type of a target road segment indicated by the target point cluster based on a difference between the first singular value and the second singular value;

[0016] Determine a target road line indicated by the target point cluster as a first target road line, and determine a first inlier rate of a point cloud corresponding to the first target road line based on an inlier rate algorithm corresponding to the road type;

[0017] When the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line.

[0018] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, wherein the processor is used to execute a computer program stored in a memory, wherein the computer program, when executed by the processor, implements the steps of any one of the road data analysis methods provided in the first aspect.

[0019] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any one of the road data analysis methods provided in the first aspect are implemented.

[0020] In a fifth aspect, an embodiment of the present application provides a vehicle, comprising: a road data analysis device as described in the second aspect, or an electronic device as described in the third aspect, or a computer-readable storage medium as described in the fourth aspect.

[0021] In a sixth aspect, an embodiment of the present application provides a computer program product, wherein the computer program product includes a computer program or instructions. When the computer program product runs on a processor, the processor executes the computer program or instructions to implement the steps of the road data analysis method as described in the first aspect.

[0022] In the seventh aspect, an embodiment of the present application provides a chip, which includes a processor, a memory and a communication interface, the communication interface is coupled to the processor, the memory is used to store programs or instructions that can be run on the processor, and the processor is used to execute the program or instructions to implement the steps of the road data analysis method as described in the first aspect.

[0023] Compared with the prior art, the technical solution provided by the embodiment of the present application has the following advantages: In the embodiment of the present application, when removing abnormal lines in the BEV road perception image, first, the road point cloud in the BEV road perception image is clustered according to the point spacing to obtain N point clusters, where N is a positive integer, and each point cluster in the N point clusters is used to indicate a different road segment, and the point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, and the target road line is a lane line or a road edge line; one point cluster in the N point clusters is used as the target point cluster, and the target point cluster is subjected to singular value decomposition to obtain a first singular value and a second singular value, and the first singular value is used to indicate the target point The variance of the point cloud distribution of the cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster; based on the difference between the first singular value and the second singular value, the road type of the target road segment indicated by the target point cluster is determined; a target road line indicated by the target point cluster is determined as the first target road line, and based on the inlier rate algorithm corresponding to the road type, the first inlier rate of the point cloud corresponding to the first target road line is determined; when the first inlier rate is less than or equal to the inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line. In this solution, the road point cloud in the BEV road perception image is clustered according to the spacing of the points to obtain N point clusters, one of the N point clusters is used as the target point cluster, and the target point cluster is subjected to singular value decomposition to obtain a first singular value and a second singular value. Based on the difference between the first singular value and the second singular value, the road type of the target road segment indicated by the target point cluster is determined, a target road line indicated by the target point cluster is determined as the first target road line, and based on the inlier rate algorithm corresponding to the road type, the first inlier rate of the point cloud corresponding to the first target road line is determined; when the first inlier rate is less than or equal to the inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line. That is, different deformed line determination methods are determined for different road types, and it is analyzed whether there are deformed lines in the road data corresponding to different road types. In this way, it can avoid determining a normal road line in the road line as a deformed line, improve the accuracy of the deformed line analysis, and after the analyzed deformed lines are eliminated, the obtained BEV road perception image can more truly and accurately reflect the local road information. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to control the vehicle more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0025] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0026] Figure 1 A schematic diagram of a road data analysis method provided in this application;

[0027] Figure 2 A schematic diagram of a flow chart of a method for determining a road type provided in this application;

[0028] Figure 3 A schematic diagram of a flow chart of a road data analysis method provided in this application;

[0029] Figure 4 A schematic diagram of a flow chart of another road data analysis method provided by the present application;

[0030] Figure 5 A schematic diagram of a road data analysis method provided in this application;

[0031] Figure 6 A schematic diagram of a flow chart of another road data analysis method provided by the present application;

[0032] Figure 7 A schematic diagram of a flow chart of another road data analysis method provided by the present application;

[0033] Figure 8 A schematic diagram of the structure of a road data analysis device provided in this application;

[0034] Fig. 9 A schematic diagram of the hardware structure of an electronic device provided in this application. DETAILED DESCRIPTION

[0035] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the scheme of the present application will be further described below. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0036] In the following description, many specific details are set forth to facilitate a full understanding of the present application, but the present application may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only part of the embodiments of the present application, rather than all of the embodiments.

[0037] The terms "first", "second", etc. in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0038] First, some nouns or terms involved in the claims and description of the present invention are explained below.

[0039] BEV (Bird's Eye View) perception: is a perception technology that observes scenes from a bird's eye view, which can provide more comprehensive and accurate environmental perception information. BEV perception has been widely used and studied in the fields of autonomous driving, intelligent transportation, logistics and distribution, because it can effectively solve the limitations of traditional monocular and binocular vision perception technologies in scene perception range, blind spots, posture changes, etc.

[0040] Clustering: It is the process of grouping data objects based on information found in the data that describes the objects and their relationships. The goal is that objects within a group are similar to each other (related), while objects in different groups are different (unrelated). The greater the similarity within a group and the greater the difference between groups, the better the clustering effect.

[0041] SVD (Singular Value Decomposition) decomposition is to decompose a matrix into three matrices multiplied by each other, where the first matrix is ​​a unit orthogonal matrix, the second matrix is ​​a diagonal matrix, and the third matrix is ​​also a unit orthogonal matrix. Specifically, given an m×n matrix A, SVD decomposition can be expressed as A=U∑V^T, where U is an m×n unit orthogonal matrix, V is an n×n unit orthogonal matrix, ∑ is a diagonal matrix, and the elements on the diagonal are the singular values ​​of A.

[0042] Random Sampling Consensus (RANSC) is an iterative algorithm used to handle clustering or classification problems of large amounts of data or data samples. The RANSC algorithm randomly selects a small number of samples from the data set for clustering or classification each time, and then updates the clustering or classification results of the entire data set based on the results of these samples. Through multiple iterations, the RANSC algorithm gradually improves the clustering or classification results and eventually obtains a relatively stable and accurate result. The RANSC algorithm is suitable for processing large-scale data sets because it can effectively reduce computational complexity and improve the efficiency of clustering or classification.

[0043] In the process of BEV perceiving static road elements, it is often affected by factors such as environmental occlusion and lighting changes, and other lines that are not lane lines, road edge lines, etc. are often mistaken for road lines, or the perceived road lines are too tortuous, resulting in large deviations in the local map established downstream, which cannot be used for vehicle control and display.

[0044] In order to obtain a BEV road perception map that can truly and accurately reflect road information, the BEV road perception map needs to be preprocessed to remove abnormal lines in the BEV road perception map.

[0045] As mentioned above, currently, the method of removing abnormal lines from road lines indicated by road data is usually to remove deformed lines from road lines. The specific method of removing deformed lines is: for each road line indicated by road data, first calculate the curvature of every three points in the road line, then accumulate and calculate the average curvature of the road line, and when the average curvature is greater than a threshold, determine that the road line is an abnormal line and remove the road line. However, the above abnormal line removal method is likely to remove normal curved lines as abnormal lines, and the accuracy of abnormal line removal is low.

[0046] In order to solve the above technical problems, the embodiments of the present application provide a road data analysis method, device, electronic device, storage medium and vehicle. The method includes: when removing abnormal lines in the BEV road perception image, first, clustering the road point cloud in the BEV road perception image according to the point spacing to obtain N point clusters, N is a positive integer, each point cluster in the N point clusters is used to indicate a different road segment, and the point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, and the target road line is a lane line or a road edge line; taking one of the N point clusters as the target point cluster, performing singular value decomposition on the target point cluster to obtain a first singular value and a second singular value, the first singular value is used to indicate the point cloud distribution of the target point cluster in the main direction of the target point cluster. The second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster; based on the difference between the first singular value and the second singular value, the road type of the target road segment indicated by the target point cluster is determined; a target road line indicated by the target point cluster is determined as a first target road line, and based on the inlier rate algorithm corresponding to the road type, the first inlier rate of the point cloud corresponding to the first target road line is determined; when the first inlier rate is less than or equal to the inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line. In this solution, the road point cloud in the BEV road perception image is clustered according to the spacing of the points to obtain N point clusters, one of the N point clusters is used as the target point cluster, and the target point cluster is subjected to singular value decomposition to obtain a first singular value and a second singular value. Based on the difference between the first singular value and the second singular value, the road type of the target road segment indicated by the target point cluster is determined, a target road line indicated by the target point cluster is determined as the first target road line, and based on the inlier rate algorithm corresponding to the road type, the first inlier rate of the point cloud corresponding to the first target road line is determined; when the first inlier rate is less than or equal to the inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line. That is, different deformed line determination methods are determined for different road types, and it is analyzed whether there are deformed lines in the road data corresponding to different road types. In this way, it can avoid determining a normal road line in the road line as a deformed line, improve the accuracy of the deformed line analysis, and after the analyzed deformed lines are eliminated, the obtained BEV road perception image can more truly and accurately reflect the local road information. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to control the vehicle more accurately.

[0047] The present application is applied to the abnormal line analysis scenario in the road data corresponding to the BEV road perception image. The electronic device in the embodiment of the present application can be a vehicle-mounted terminal, or a mobile phone, notebook, computer, etc. that is connected to the vehicle for communication. The specific details can be determined based on actual conditions and are not limited here.

[0048] The technical solution of the present application is explained in detail below through several exemplary embodiments.

[0049] Figure 1 A schematic diagram of a road data analysis method provided in this application, such as Figure 1 As shown, the road data analysis method may include the following steps S101 to S105.

[0050] S101. Cluster the road point cloud in the BEV road perception image according to the point spacing to obtain N point clusters, where N is a positive integer, and each point cluster in the N point clusters is used to indicate a different road segment. The point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, and the target road line is a lane line or a road edge line.

[0051] Among them, the road point cloud in the BEV road perception image can be a 2D intersection point cloud, and the 2D intersection point cloud can be split into roads using a clustering algorithm.

[0052] N is a positive integer. For example, when N is equal to 1, the road scene indicated by the BEV road perception image is determined to be a non-intersection scene; when N is equal to 2, the road scene of the BEV road perception image is determined to be a two-way intersection scene; when N is equal to 3, the road scene of the BEV road perception image is determined to be a three-way intersection scene; when N is equal to 4, the road scene of the BEV road perception image is determined to be a four-way intersection scene.

[0053] Among them, non-intersection scenes refer to straight road or curve scenes, two-way intersections include intersections in two directions, such as L-shaped intersections, three-way intersections include intersections in three directions, such as three-way intersections, T-junctions, etc., and four-way intersection scenes include intersections in four directions, such as crossroads, X-shaped intersections, etc.

[0054] The embodiment of the present application does not impose any specific limitation on the clustering method of the above-mentioned road point cloud.

[0055] In some embodiments, the road point cloud in the BEV road perception image is clustered according to the point spacing through the following steps S101A1 to S101A2 to obtain N point clusters.

[0056] S101A1. Determine the cluster spacing.

[0057] S101A2. Based on the clustering distance, the road point cloud in the BEV road perception image is clustered according to the distance between the points to obtain N point clusters.

[0058] Among them, the above-mentioned cluster spacing in the embodiment of the present application can refer to the distance between different intersections and sections. The distance can be the Euclidean distance between different intersections and sections, or other distances, such as cosine distance, Manhattan distance, etc. The specific distance can be determined according to actual conditions and is not limited here.

[0059] In a possible implementation, the above S101A2 "based on the clustering distance, Euclidean clustering is performed on the road point cloud in the BEV road perception image to obtain N point clusters" includes the following S101A2-11 to S101A2-15.

[0060] S101A2-11. Select a point P10 from the road point cloud in the BEV road perception image.

[0061] S101A2-12. Determine the distance from P10 to M points other than P10 in the road point cloud. When the distance is less than or equal to the clustering distance, determine the point and P10 as points in the same point cluster C1. When the distance is greater than the clustering distance, determine the point and P10 as points in different point clusters.

[0062] S101A2-13. Traverse all unvisited points in C1 and repeat step S101A2-12.

[0063] S101A2-14. When all points in C1 are visited, the C1 clustering ends.

[0064] S101A2-15. Select a point P20 from the road point cloud, where P20 is any point in the road point cloud except C1, and iterate the above steps S101A2-11 to S101A2-14 repeatedly until all points in the road point cloud have been visited, then terminate all clustering and obtain N point clusters.

[0065] Other embodiments of the present invention are further optimized based on the above embodiments. In some embodiments, the above S101A2 "Based on the clustering distance, the road point cloud in the BEV road perception image is subjected to Euclidean clustering to obtain N point clusters" includes the following steps S101A2-21 to S101A2-23.

[0066] S101A2-21. Cluster the road point cloud based on the clustering distance to obtain M point clusters, where M is an integer greater than or equal to N.

[0067] S101A2-22. Determine the number of point clouds in each of the M point clusters.

[0068] S101A2-23. Determine the point clusters whose point cloud quantity is greater than or equal to the quantity threshold among the M point clusters as N point clusters.

[0069] S102. Take one of the N point clusters as the target point cluster, perform singular value decomposition on the target point cluster, and obtain a first singular value and a second singular value, wherein the first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction.

[0070] The main direction of the point cluster may be the driving direction of the road segment. The main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster.

[0071] Among them, singular value decomposition is also called SVD decomposition. By performing SVD decomposition on the target point cluster, a singular vector V1 and a singular vector V2, as well as a first singular value corresponding to the singular vector V1 and a second singular value corresponding to the singular vector V2 can be obtained. The singular vector V1 is used to indicate the main direction of the target point cluster, and the singular vector V2 is used to indicate the direction of the target point cluster perpendicular to the main direction. The first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the direction perpendicular to the main direction.

[0072] It should be understood that the main direction of the target point cluster is usually the direction in which the car is traveling, and the direction perpendicular to the main direction is usually the road width direction.

[0073] S103: Determine a road type of a target road segment indicated by the target point cluster based on a difference between the first singular value and the second singular value.

[0074] The embodiment of the present application does not limit the specific method of “determining the road type of the target road segment indicated by the target point cluster based on the difference between the first singular value and the second singular value”.

[0075] Figure 2 It is a flowchart of a method for determining a road type provided in an embodiment of the present application.

[0076] In some embodiments, Figure 2 As shown, the above S103 is specifically based on the following steps S103A1 to S103A4 to determine the road type of the target road segment indicated by the target point cluster.

[0077] S103A1. Determine a first parameter according to a first singular value and a second singular value, where the first parameter is used to indicate a difference between the first singular value and the second singular value.

[0078] In some embodiments, "determining a first parameter according to the first singular value and the second singular value" in S103A1 includes: determining the ratio of the first singular value to the second singular value as the first parameter; or determining the difference between the first singular value and the second singular value as the first parameter.

[0079] The embodiment of the present application provides two methods for determining the first parameter. The first method is to determine the ratio of the first singular value to the second singular value as the first parameter, and the second method is to determine the difference between the first singular value and the second singular value as the first parameter. Among them, the ratio of the first singular value to the second singular value, and the difference between the first singular value and the second singular value can clearly reflect the difference between the first singular value and the second singular value. In this way, the process of determining the first parameter is simple and clear, and at the same time more efficient.

[0080] In some other embodiments, "determining a first parameter according to the first singular value and the second singular value" in S103A1 includes: comparing the first singular value with the second singular value to obtain a first ratio, and determining the product of the first ratio and a preset value as the first parameter.

[0081] S103A2: When the first parameter is greater than or equal to the first threshold, determine that the road type of the target road segment indicated by the target point cluster is a straight road.

[0082] S103A3: When the first parameter is less than the first threshold value and greater than the second threshold value, determine that the road type of the target road segment indicated by the target point cluster is a small curvature curve.

[0083] S103A4: When the first parameter is less than or equal to the second threshold, determine that the road type of the target road segment indicated by the target point cluster is a large curvature curve.

[0084] The first threshold is greater than the second threshold, and the curvature of the small curvature curve is smaller than the curvature of the large curvature curve.

[0085] It should be understood that, first, before obtaining the BEV road perception image, it is necessary to collect the road information of X meters in front and behind the vehicle, and the road information of Y meters to the left and right of the vehicle through the on-board camera, and then obtain the BEV road perception image based on the road information obtained by the on-board camera. When actually collecting road information, Y does not exceed the road width, while X is much larger than the road width. Therefore, the point cloud distribution in the main direction of the road perception image is relatively discrete, while the point cloud distribution in the direction perpendicular to the main direction is relatively concentrated. Therefore, the variance of the point cloud distribution in the main direction is large, while the variance of the point cloud distribution in the direction perpendicular to the main direction is small. Further, for a straight road, the difference between the variance of the point cloud distribution in the main direction and the variance of the point cloud distribution in the direction perpendicular to the main direction is the largest, for a small curvature curve, the difference between the variance of the point cloud distribution in the main direction and the variance of the point cloud distribution in the direction perpendicular to the main direction is second, and for a large curvature curve, the difference between the variance of the point cloud distribution in the main direction and the variance of the point cloud distribution in the direction perpendicular to the main direction is the smallest, therefore, according to the first parameter, it can be determined that the target road segment indicated by the target point cluster is a straight road or a small curvature curve or a large curvature curve. The larger the first parameter, the closer the road type is to a straight road.

[0086] Other embodiments of the present invention are further optimized based on the above embodiments. In some embodiments, the above S103 is specifically based on the following steps S103B1 to S103B5 to determine the road type of the target road segment indicated by the target point cluster.

[0087] S103B1. Determine a first parameter according to the first singular value and the second singular value, where the first parameter is used to indicate a difference between the first singular value and the second singular value.

[0088] S103B2: When the first parameter is greater than or equal to the first threshold, determine that the road type of the target road segment indicated by the target point cluster is a straight road.

[0089] S103B3: When the first parameter is less than the first threshold value and greater than the second threshold value, determine that the road type of the target road segment indicated by the target point cluster is a small curvature curve.

[0090] S103B4: When the first parameter is less than or equal to the second threshold and greater than the third threshold, determine that the road type of the target road segment indicated by the target point cluster is a medium curvature curve.

[0091] S103B5. When the first parameter is less than or equal to the third threshold, determine that the road type of the target road segment indicated by the target point cluster is a large curvature curve.

[0092] Among them, the first threshold is greater than the second threshold, the second threshold is greater than the third threshold, the curvature of the small curvature curve is smaller than the curvature of the medium curvature curve, and the curvature of the medium curvature curve is smaller than the curvature of the large curvature curve.

[0093] S104: Determine a target road line indicated by the target point cluster as a first target road line, and determine a first inlier rate of a point cloud corresponding to the first target road line based on an inlier rate algorithm corresponding to the road type.

[0094] In some embodiments of the present application, in the above-mentioned step S104, based on the inlier rate algorithm corresponding to the road type, determining the first inlier rate of the point cloud corresponding to the first target road line can be as follows: determining the corresponding preset number according to the road type, selecting a preset number of reference points from the point cloud of the first target road line to generate a reference line corresponding to the road type, determining the number of points that meet the target condition from other points in the point cloud of the first target road line, the target condition being that the distance from the point to the reference line is less than or equal to the reference threshold corresponding to the road type, and determining the first inlier rate based on the ratio of the number of points that meet the target condition to the total number of point clouds corresponding to the first target road line.

[0095] In some embodiments of the present application, the first inlier rate is determined based on the ratio of the number of points that meet the target condition to the total number of point clouds corresponding to the first target road line. The first inlier rate can be determined by the ratio of the number of points that meet the target condition to the total number of point clouds corresponding to the first target road line; or the ratio of the number of points that meet the target condition to the total number of point clouds corresponding to the first target road line can be determined as the inlier rate corresponding to the reference line, and then the above-mentioned steps of determining the inlier rate corresponding to the reference line are executed in a loop, and the inlier rate corresponding to the reference line with the largest number of points that meet the target condition is determined as the first inlier rate.

[0096] The inlier rate algorithm corresponding to the road type may be a random sampling consensus (Ransac) algorithm corresponding to the road type.

[0097] It should be understood that two different points are defined in the Ransac algorithm: in-game points and out-game points. In-game points refer to points that conform to the algorithm model corresponding to the Ransac algorithm, and correspondingly, out-game points refer to points that do not conform to the algorithm model corresponding to the Ransac algorithm. The in-game point rate is equal to the ratio of the number of in-game points to the number of all points.

[0098] Exemplarily, the above-mentioned inlier rate algorithm based on the road type corresponds to determine the first inlier rate of the point cloud corresponding to the first target road line, including: when the above-mentioned road type is a straight road, through the following steps S104A1 to S104A5, determine the first inlier rate of the point cloud corresponding to the first target road line.

[0099] S104A1. Randomly select 2 points from the point cloud of the first target road line.

[0100] S104A2. Calculate the equation of the straight line from these two points.

[0101] S104A3. Calculate the distances from other points in the point cloud of the first target road line to the straight line, and determine the number of points that meet the target condition (the distance from the point to the straight line is less than or equal to the reference threshold corresponding to the straight road).

[0102] S104A4. Determine the ratio of the number of points that meet the target condition to the total number of point clouds of the first target road line as the inner point rate corresponding to the straight line.

[0103] S104A5, loop through steps 1-4, record the straight line with the largest number of points satisfying the target condition, and determine the inlier rate corresponding to the straight line as the first inlier rate.

[0104] Exemplarily, the above-mentioned inlier rate algorithm based on the road type corresponds to determine the first inlier rate of the point cloud corresponding to the first target road line, including: when the above-mentioned road type is a small curvature curve, through the following steps S104B1 to S104B5, determine the first inlier rate of the point cloud corresponding to the first target road line.

[0105] S104B1. Randomly select 3 points from the point cloud of the first target road line.

[0106] S104B2. Calculate the curve equation from these three points.

[0107] S104B3. Calculate the distances from other points in the point cloud of the first target road line to the curve, and determine the number of points that meet the target condition (the distance from the point to the curve is less than or equal to the reference threshold corresponding to the small curvature curve).

[0108] S104B4. Determine the ratio of the number of points that meet the target condition to the total number of point clouds of the first target road line as the inner point rate corresponding to the curve.

[0109] S104B5, loop through steps 1-4, record the curve with the largest number of points satisfying the target condition, and determine the inlier rate corresponding to the curve as the first inlier rate.

[0110] When the road type is a curve with a large curvature, the method for determining the first inlier rate of the point cloud corresponding to the first target road line can refer to the above steps S104B1 to S104B5, which will not be repeated here.

[0111] S105: When the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type, determine that the first target road line is a deformed line.

[0112] It can be understood that when the first inlier rate is greater than the inlier rate threshold corresponding to the road type, it is determined that the first target road line is not a deformed line.

[0113] For each target road line in the target point cluster, the above steps S104 and S105 are repeatedly executed to analyze the deformed lines included in the target point cluster, and then the point cloud corresponding to the deformed line in the target road line indicated by the target point cluster can be removed from the BEV road perception image.

[0114] The deformed line refers to a road line whose deformation degree is greater than a certain threshold. For example, if the bending degree of a lane line is greater than a first threshold, the lane line is determined to be a deformed line.

[0115] Exemplarily, the inlier rate threshold corresponding to a curve with small curvature is determined to be lambda5, the inlier rate threshold corresponding to a curve with large curvature is determined to be lambda6, and the inlier rate threshold corresponding to a straight road is determined to be lambda7.

[0116] When the above-mentioned road type is a straight road, determine whether the first inlier rate corresponding to the first target road line is less than or equal to lambda7. When the first inlier rate is less than or equal to lambda7, determine that the first target road line is a deformed line, so as to remove the point cloud corresponding to the deformed line in the target road line indicated by the target point cluster from the BEV road perception image.

[0117] When the above-mentioned road type is a small curvature curve, determine whether the first inlier rate corresponding to the first target road line is less than or equal to lambda5. When the first inlier rate is less than or equal to lambda5, determine that the first target road line is a deformed line, so as to remove the point cloud corresponding to the deformed line in the target road line indicated by the target point cluster from the BEV road perception image.

[0118] When the above-mentioned road type is a large curvature curve, it is determined whether the first inlier rate corresponding to the first target road line is less than or equal to lambda6. When the first inlier rate is less than or equal to lambda6, the first target road line is determined to be a deformed line, so as to remove the point cloud corresponding to the deformed line in the target road line indicated by the target point cluster from the BEV road perception image.

[0119] In the embodiment of the present application, the road point cloud in the BEV road perception image is clustered to obtain N point clusters; one of the N point clusters is used as the target point cluster, and the target point cluster is subjected to singular value decomposition to obtain a first singular value and a second singular value; based on the difference between the first singular value and the second singular value, the road type of the target road segment indicated by the target point cluster is determined; a target road line indicated by the target point cluster is determined as the first target road line, and based on the Ransac algorithm corresponding to the road type, the first inlier rate of the point cloud corresponding to the first target road line is determined; when the first inlier rate is less than or equal to the inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line, so that the point cloud corresponding to the deformed line in the target road line indicated by the target point cluster is removed from the BEV road perception image. That is, different deformed line determination methods are determined for different road types, so that normal road lines in road lines can be avoided from being removed as abnormal lines, the accuracy of deformed line removal is improved, and the obtained BEV road perception image can more truly and accurately reflect local road information. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to control the vehicle more accurately.

[0120] In addition, "determining the road type of the target road segment indicated by the target point cluster based on the difference between the first singular value and the second singular value" can be specifically achieved in the following manner: determining a first parameter according to the first singular value and the second singular value, the first parameter being used to indicate the difference between the first singular value and the second singular value; determining that the road type of the target road segment indicated by the target point cluster is a straight road when the first parameter is greater than or equal to the first threshold; determining that the road type of the target road segment indicated by the target point cluster is a small curvature curve when the first parameter is less than the first threshold and greater than the second threshold; determining that the road type of the target road segment indicated by the target point cluster is a large curvature curve when the first parameter is less than or equal to the second threshold. In this way, the road types are divided into straight roads, large curvature curves, and small curvature curves. Different deformed line determination methods are formulated for straight roads, large curvature curves, and small curvature curves, respectively, so as to avoid removing normal road lines in the road lines as abnormal lines, thereby improving the accuracy of removing deformed lines, and the obtained BEV road perception image can more truly and accurately reflect local road information. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to control the vehicle more accurately.

[0121] The above-mentioned embodiments specifically illustrate the method of removing deformed lines in lane lines and road edge lines in the present application. In other embodiments, the present application not only removes deformed lines in lane lines and road edge lines, but also removes outlier lines in lane lines and road edge lines. The following embodiments specifically illustrate the scheme for removing outlier lines in lane lines and road edge lines.

[0122] Among them, the deformed line refers to the line within the road range, and the outlier line refers to the line not within the road range.

[0123] It should be noted that in the process of BEV perceiving static road elements, it is affected by factors such as environmental occlusion and lighting changes, and other lines that are not lane lines, road edge lines, etc. are often mistaken for road lines, resulting in large deviations in the local map established downstream, which cannot be used for vehicle control and display.

[0124] Normally, these lines mistakenly identified as “road lines” usually deviate from the main direction of the road or are far away from normal road lines. In the embodiment of the present application, “road lines” in the BEV road perception image whose angle with the main direction of the road exceeds a certain threshold or whose distance from other road lines exceeds a certain threshold are defined as “outlier lines”.

[0125] At present, the scheme for removing abnormal lines in the road lines indicated by the road data is: for each road line indicated by the road data, the curvature of every three points in the road line is first calculated, and then the average curvature of the road line is accumulated and calculated. When the average curvature is greater than a threshold, the road line is determined to be an abnormal line and is removed. However, this abnormal line removal scheme cannot remove outliers with a relatively small degree of distortion.

[0126] In order to solve the above technical problems, the embodiment of the present application provides the following solution to remove outlier lines that are far away from normal road lines in the BEV road perception image, thereby improving the accuracy of map production.

[0127] Figure 3 A schematic diagram of a flow chart of a road data analysis method provided in an embodiment of the present application. Figure 3 As shown, the road data analysis method may include the following steps S301 to S303.

[0128] S301: Determine a target road line indicated by a target point cluster as a second target road line, and determine the centroid of the point cloud corresponding to the second target road line.

[0129] Exemplarily, the average value of the coordinate values ​​of all point clouds in the second target road is the coordinate value of the centroid.

[0130] S302: Determine a first distance between the centroid of the point cloud corresponding to the second target road line and the cluster center of the target point cluster.

[0131] Exemplarily, the coordinates corresponding to the above cluster center are the average values ​​of all points in the target point cluster.

[0132] S303: When the first distance is greater than or equal to a first distance threshold, determine that the second target road line is an outlier line.

[0133] For each target road line in the target point cluster, the above steps S301 to S303 are repeatedly executed to analyze the outlier lines included in the target point cluster. The outlier lines are also abnormal lines in the road data. Then, the point cloud corresponding to the outlier lines in the target road line indicated by the target point cluster can be removed from the BEV road perception image.

[0134] In an embodiment of the present application, a target road line indicated by a target point cluster is determined as a second target road line, and the centroid of the point cloud corresponding to the second target road line is determined; the first distance between the centroid of the point cloud corresponding to the second target road line and the cluster center of the target point cluster is determined; when the first distance is greater than or equal to the first distance threshold, the second target road line is determined as an outlier line, so that the point cloud corresponding to the outlier line in the target road line indicated by the target point cluster is removed from the BEV road perception image. In this way, outliers that deviate from other road lines can be determined from the road line, and the point clouds corresponding to the outliers that deviate from other road lines can be removed from the BEV road perception image. In this way, no matter how large or small the degree of distortion of the outliers in the BEV road perception image is, they can be removed, thereby improving the accuracy of outlier removal. The obtained BEV road perception image can reflect local road information more realistically and accurately. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to more accurately control the vehicle.

[0135] In the above-provided outlier line elimination scheme, "a target road line indicated by the target point cluster is determined as the second target road line, and the centroid of the point cloud corresponding to the second target road line is determined; the first distance between the centroid of the point cloud corresponding to the second target road line and the cluster center of the target point cluster is determined; when the first distance is greater than or equal to the first distance threshold, the second target road line is determined as an outlier line, so as to eliminate the point cloud corresponding to the outlier line in the target road line indicated by the target point cluster from the BEV road perception image." That is to say, the above scheme can determine the outliers that are far from the normal road lines from the road lines, and eliminate the point clouds corresponding to the outliers from the BEV road perception image. However, no corresponding solution is provided for the outliers that are not far from the normal road lines but deviate from the main direction of the road.

[0136] Other embodiments of the present invention are further optimized based on the above embodiments. In some embodiments, another road data analysis method is provided, which can remove outliers that are far from normal road lines and outliers that deviate from the main direction of the road to further improve the accuracy of map making.

[0137] Figure 4 A schematic diagram of a flow chart of another road data analysis method provided in an embodiment of the present application. Figure 4 As shown, the road data analysis method may include the following steps S401 to S405.

[0138] S401: Determine a target road line indicated by a target point cluster as a second target road line, and determine the centroid of the point cloud corresponding to the second target road line.

[0139] S402: Determine a first distance between the centroid of the point cloud corresponding to the second target road line and the cluster center of the target point cluster.

[0140] S403: When the first distance is greater than or equal to a first distance threshold, determine that the second target road line is an outlier line.

[0141] S404: When the first distance is less than a first distance threshold, perform singular value decomposition on the second target road line to obtain a first singular vector, where the first singular vector is used to indicate a main direction of the second target road line.

[0142] The singular value decomposition is also called SVD decomposition. By performing SVD decomposition on the second target road line, a first singular vector can be obtained. The first singular vector is used to indicate the main direction of the second target road line.

[0143] S405: When the angle between the main direction of the second target road line and the main direction of the target point cluster is greater than or equal to the angle threshold, determine that the second target road line is an outlier line.

[0144] For each target road line in the target point cluster, the above steps S401 to S405 are repeatedly executed to analyze the outlier lines included in the target point cluster. The outlier lines are also abnormal lines in the road data. Then, the point cloud corresponding to the outlier lines in the target road line indicated by the target point cluster can be removed from the BEV road perception image.

[0145] In an embodiment of the present application, a target road line indicated by a target point cluster is determined as a second target road line, and the centroid of the point cloud corresponding to the second target road line is determined; a first distance between the centroid of the point cloud corresponding to the second target road line and the cluster center of the target point cluster is determined; when the first distance is greater than or equal to a first distance threshold, the second target road line is determined to be an outlier line, so that the point cloud corresponding to the outlier line in the target road line indicated by the target point cluster is eliminated from the BEV road perception image; when the first distance is less than the first distance threshold, the second target road line is subjected to singular value decomposition to obtain a first singular vector, and the first singular vector is used to indicate the main direction of the second target road line; when the angle between the main direction of the second target road line and the main direction of the target point cluster is greater than or equal to the angle threshold, the second target road line is determined to be an outlier line, so that the point cloud corresponding to the outlier line in the target road line indicated by the target point cluster is eliminated from the BEV road perception image. In this way, outliers that are far from the normal road lines can be removed from the BEV road perception image, and outliers that deviate from the main direction of the road can also be removed from the BEV road perception image. In this way, regardless of the degree of distortion of the outliers in the BEV road perception image, and regardless of the type of outliers, they can be removed, further improving the accuracy of outlier removal, and the obtained BEV road perception image can more realistically and accurately reflect the local road information. In this way, when establishing a local map based on the BEV road perception image, more accurate map information can be obtained to more accurately control the vehicle.

[0146] It should be noted that the method of removing outliers from a road line as shown in steps S301 to S303, or the method of removing outliers from a road line as shown in steps S401 to S405, can be performed before step S102, that is, the present application can first remove outliers from a road line, and then remove deformed lines from a road line, so that the second target road line is any one of the at least one target road line indicated by the target point cluster, if the second target road line is an outlier, then the first target road line is any one of the other target road lines except the second target road line in the at least one target road line; if the second target road line is not an outlier, then the first target road line is any one of the at least one target road line (including the second target road line). At the same time, the method of removing outliers from a road line can also be performed after S105, that is, the present application can first remove deformed lines from a road line, and then remove outliers from the road. In this way, the first target road line is any one of the at least one target road line indicated by the target point cluster. If the first target road line is a deformed line, the second target road line is any one of the other target road lines in the at least one target road line except the first target road line. If the first target road line is not a deformed line, the second target road line is any one of the at least one target road line (including the first target road line).

[0147] In addition to sensing lane lines and road edge lines on the road, BEV environmental perception technology can also sense other static elements on the road, such as stop lines, ground arrows, ground text, etc.

[0148] The above-mentioned embodiments specifically illustrate the method of the present application for removing abnormal lines (including deformed lines and outlier lines) in lane lines and road edge lines in BEV road perception images. In other embodiments, the present application can not only remove abnormal lines in lane lines and road edge lines in BEV road perception images, but also remove abnormal lines in stop lines in BEV road perception images. The following embodiments specifically illustrate the scheme for removing abnormal lines in stop lines in BEV road perception images.

[0149] In some embodiments, the point cloud included in each of the N point clusters is further used to indicate at least one stop line. Figure 5 A schematic diagram of a road data analysis method provided in an embodiment of the present application. Figure 5 As shown, the method also includes the following steps:

[0150] S501: Determine a stop line indicated by a target point cluster as a first target stop line, and determine a second inlier rate of a point cloud corresponding to the first target stop line based on an inlier rate algorithm corresponding to the stop line.

[0151] Exemplarily, the inlier rate algorithm corresponding to the stop line may be a Ransac algorithm corresponding to the stop line.

[0152] Among them, the stop line is usually a straight line. Therefore, the specific process of determining the second inlier rate of the point cloud corresponding to the first target stop line based on the inlier rate algorithm corresponding to the stop line can refer to the above steps S104A1 to S104A5 and will not be repeated here.

[0153] S502: When the second inlier rate is less than or equal to the inlier rate threshold corresponding to the stop line, determine that the first target stop line is a deformed line.

[0154] Among them, for each stop line indicated by the target point cluster, the above steps S501 to S502 are repeatedly executed to analyze the outlier lines in the stop lines indicated by the target point cluster. The outlier lines are also abnormal lines in the road data. Then, the point cloud corresponding to the outlier lines in the stop lines indicated by the target point cluster can be removed from the BEV road perception image.

[0155] In an embodiment of the present application, a stop line indicated by a target point cluster is determined as a first target stop line, and based on an inlier rate algorithm corresponding to the stop line, a second inlier rate of the point cloud corresponding to the first target stop line is determined; when the second inlier rate is less than or equal to the inlier rate threshold corresponding to the stop line, the first target stop line is determined to be a deformed line, so that the point cloud corresponding to the deformed line in the stop line indicated by the target point cluster is removed from the BEV road perception image. In this way, the point cloud corresponding to the deformed line in the stop line can be removed from the BEV road perception image, so that the obtained BEV road perception image can reflect the local road information more realistically and accurately. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to more accurately control the vehicle.

[0156] It should be noted that in the process of BEV perceiving static road elements, due to factors such as environmental occlusion and lighting changes, other lines that are not stop lines are often determined as stop lines, resulting in large deviations in the local map established downstream, which cannot be used for vehicle control and display.

[0157] Normally, the lines mistaken for "stop lines" are far away from normal stop lines, or deviate from the normal stop line direction. For example, the normal stop line direction should be at a 90-degree angle to the main direction of the road. If the angle between a stop line and the main direction of the road is much greater than or much less than a certain threshold, it should be considered an outlier. In the embodiment of the present application, a "stop line" in the BEV road perception image whose angle with the main direction of the road exceeds a certain angle range, or whose distance from other stop lines exceeds a certain threshold, is defined as an "outlier line".

[0158] At present, the scheme for removing abnormal lines in the road lines indicated by the road data is: for each road line indicated by the road data, the curvature of every three points in the road line is first calculated, and then the average curvature of the road line is accumulated and calculated. When the average curvature is greater than a threshold, the road line is determined to be an abnormal line and is removed. However, this abnormal line removal scheme cannot remove outliers with a relatively small degree of distortion.

[0159] In order to solve the above technical problems, the embodiment of the present application provides the following solution to remove outlier lines that are far away from the normal stop line in the BEV road perception image, thereby improving the accuracy of map production.

[0160] Figure 6 A schematic diagram of a flow chart of another road data analysis method provided in an embodiment of the present application. Figure 6 As shown, the road data analysis method may include the following steps S601 to S603:

[0161] S601: Determine a stop line indicated by a target point cluster as a second target stop line, and determine the centroid of the point cloud corresponding to the second target stop line.

[0162] S602: Determine a second distance between the centroid of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster.

[0163] S603: When the second distance is greater than or equal to a second distance threshold, determine that the second target stop line is an outlier line.

[0164] Among them, for each stop line indicated by the target point cluster, the above steps S501 to S502 are repeatedly executed to analyze the outlier lines in the stop lines indicated by the target point cluster. The outlier lines are also abnormal lines in the road data. Then, the point cloud corresponding to the outlier lines in the stop lines indicated by the target point cluster can be removed from the BEV road perception image.

[0165] In the embodiment of the present application, a stop line indicated by the target point cluster is determined as the second target stop line, and the centroid of the point cloud corresponding to the second target stop line is determined; the second distance between the centroid of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster is determined; when the second distance is greater than or equal to the second distance threshold, the second target stop line is determined as an outlier line, so that the point cloud corresponding to the outlier line in the stop line indicated by the target point cluster is removed from the BEV road perception image. In this way, the outlier line that deviates from other stop lines can be determined from the stop line, and the point cloud corresponding to the outlier line that deviates from other stop lines can be removed from the BEV road perception image. In this way, regardless of the degree of distortion of the outlier line in the BEV road perception image, it can be removed, thereby improving the accuracy of outlier line removal. The obtained BEV road perception image can reflect local road information more realistically and accurately. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to control the vehicle more accurately.

[0166] In the above-provided outlier line elimination scheme, "a stop line indicated by the target point cluster is determined as the second target stop line, and the centroid of the point cloud corresponding to the second target stop line is determined; the second distance between the centroid of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster is determined; when the second distance is greater than or equal to the second distance threshold, the second target stop line is determined as an outlier line, so as to eliminate the point cloud corresponding to the outlier line in the stop line indicated by the target point cluster from the BEV road perception image." That is to say, the above scheme can determine the stop line that is far from the normal stop line from the stop line, and eliminate the point cloud corresponding to the stop line from the BEV road perception image. However, no corresponding solution is provided for the stop line that is not far from the normal stop line, but the angle with the main direction of the road is much greater than or much less than a certain threshold.

[0167] Other embodiments of the present invention are further optimized on the basis of the above embodiments. In some embodiments, another road data analysis method is provided, which can remove the point clouds corresponding to the stop lines that are far from the normal stop lines in the BEV road perception image, and the point clouds corresponding to the stop lines whose angles with the main direction of the road are much greater than or much less than a certain threshold, so as to further improve the accuracy of map production.

[0168] Figure 7 A schematic diagram of a flow chart of another road data analysis method provided in an embodiment of the present application. Figure 7 As shown, the road data analysis method may include the following steps S701 to S405:

[0169] S701: Determine a stop line indicated by the target point cluster as a second target stop line, and determine the centroid of the point cloud corresponding to the second target stop line.

[0170] S702: Determine a second distance between the centroid of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster.

[0171] S703: When the second distance is greater than or equal to a second distance threshold, determine that the second target stop line is an outlier line.

[0172] Among them, for each stop line indicated by the target point cluster, the above steps S501 to S502 are repeatedly executed to analyze the outlier lines in the stop lines indicated by the target point cluster. The outlier lines are also abnormal lines in the road data. Then, the point cloud corresponding to the outlier lines in the stop lines indicated by the target point cluster can be removed from the BEV road perception image.

[0173] S704. When the second distance is less than a second distance threshold, perform singular value decomposition on the point cloud corresponding to the second target stop line to obtain a second singular vector, where the second singular vector is used to indicate a main direction of the second target stop line.

[0174] S705: When the angle between the main direction of the second target stop line and the main direction of the target point cluster is not within a preset angle range, determine that the second target stop line is an outlier line.

[0175] Among them, for each stop line indicated by the target point cluster, the above steps S501 to S502 are repeatedly executed to analyze the outlier lines in the stop lines indicated by the target point cluster. The outlier lines are also abnormal lines in the road data. Then, the point cloud corresponding to the outlier lines in the stop lines indicated by the target point cluster can be removed from the BEV road perception image.

[0176] In an embodiment of the present application, a stop line indicated by a target point cluster is determined as a second target stop line, and the centroid of the point cloud corresponding to the second target stop line is determined; a second distance between the centroid of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster is determined; when the second distance is greater than or equal to a second distance threshold, the second target stop line is determined to be an outlier line, so that the point cloud corresponding to the outlier line in the stop line indicated by the target point cluster is eliminated from the BEV road perception image; when the second distance is less than the second distance threshold, the point cloud corresponding to the second target stop line is subjected to singular value decomposition to obtain a second singular vector, and the second singular vector is used to indicate the main direction of the second target stop line; when the angle between the main direction of the second target stop line and the main direction of the target point cluster is not within a preset angle range, the second target stop line is determined to be an outlier line, so that the point cloud corresponding to the outlier line in the stop line indicated by the target point cluster is eliminated from the BEV road perception image. In this way, the stop lines that are far from the normal stop lines can be removed from the BEV road perception image, and the point clouds corresponding to the stop lines whose angles with the main direction of the road are much larger or much smaller than a certain threshold can be removed from the BEV road perception image. In this way, regardless of the degree of distortion of the outlier lines in the BEV road perception image, and regardless of the type of outlier lines, they can be removed, further improving the accuracy of outlier line removal, and the obtained BEV road perception image can more realistically and accurately reflect the local road information. In this way, when establishing a local map based on the BEV road perception image, more accurate map information can be obtained to more accurately control the vehicle.

[0177] In summary, the present application provides a road data analysis method, including clustering a road point cloud in a BEV road perception image according to the spacing of points to obtain N point clusters; taking one of the N point clusters as a target point cluster, performing singular value decomposition on the target point cluster to obtain a first singular value and a second singular value; determining the road type of a target road segment indicated by the target point cluster based on the difference between the first singular value and the second singular value; determining a target road line indicated by the target point cluster as a first target road line, and determining a first inlier rate of a point cloud corresponding to the first target road line based on a Ransac algorithm corresponding to the road type; and determining the first target road line as a deformed line when the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type, so as to remove the point cloud corresponding to the deformed line in the target road line indicated by the target point cluster from the BEV road perception image. That is, different deformed line determination methods are determined for different road types, so that normal road lines in road lines can be avoided from being removed as abnormal lines, and the accuracy of abnormal line removal can be improved, and the obtained BEV road perception image can more truly and accurately reflect local road information. In this way, when a local map is established based on the BEV road perception image, more accurate map information can be obtained to control the vehicle more accurately.

[0178] The present application also provides a road data analysis device, Figure 8 A schematic diagram of the structure of a road data analysis device provided in this application, such as Figure 8 As shown, the road data analysis device 80 includes:

[0179] A clustering unit 81 is used to cluster the road point cloud in the BEV road perception image according to the point spacing to obtain N point clusters, where N is a positive integer, each point cluster in the N point clusters is used to indicate a different road segment, and the point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, where the target road line is a lane line or a road edge line;

[0180] A singular value decomposition unit 82 is used to take one of the N point clusters as a target point cluster, perform singular value decomposition on the target point cluster, and obtain a first singular value and a second singular value, wherein the first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster;

[0181] The determination unit 83 is used to determine the road type of the target road segment indicated by the target point cluster based on the difference between the first singular value and the second singular value; determine a target road line indicated by the target point cluster as a first target road line, and determine a first inlier rate of the point cloud corresponding to the first target road line based on an inlier rate algorithm corresponding to the road type; and determine that the first target road line is a deformed line when the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type.

[0182] In some embodiments, the determination unit 83 is specifically used to determine a first parameter based on the first singular value and the second singular value, the first parameter being used to indicate the difference between the first singular value and the second singular value; when the first parameter is greater than or equal to a first threshold, the road type of the target road segment indicated by the target point cluster is determined to be a straight road; when the first parameter is less than the first threshold and greater than the second threshold, the road type of the target road segment indicated by the target point cluster is determined to be a small curvature curve; when the first parameter is less than or equal to the second threshold, the road type of the target road segment indicated by the target point cluster is determined to be a large curvature curve; the first threshold is greater than the second threshold, and the curvature of the small curvature curve is less than the curvature of the large curvature curve.

[0183] In some embodiments, the determination unit 83 is specifically configured to determine the ratio of the first singular value to the second singular value as the first parameter; or determine the difference between the first singular value and the second singular value as the first parameter.

[0184] In some embodiments, the determination unit 83 is specifically used to determine the corresponding preset number according to the road type; generate a reference line corresponding to the road type based on a preset number of reference points randomly selected from the point cloud of the first target road line; determine the number of points that meet the target condition from other points in the point cloud of the first target road line, and the target condition is that the distance from the point to the reference line is less than or equal to the reference threshold corresponding to the road type; determine the first inner point rate based on the ratio of the number of points that meet the target condition to the total number of point clouds corresponding to the first target road line.

[0185] In some embodiments, the determination unit 83 is also used to determine a target road line indicated by the target point cluster as a second target road line, determine the center of mass of the point cloud corresponding to the second target road line; determine the first distance between the center of mass of the point cloud corresponding to the second target road line and the cluster center of the target point cluster; and when the first distance is greater than or equal to a first distance threshold, determine the second target road line as an outlier line.

[0186] In some embodiments, the singular value decomposition unit 82 is further used to perform singular value decomposition on the second target road line when the first distance is less than the first distance threshold to obtain a first singular vector, where the first singular vector is used to indicate a main direction of the second target road line;

[0187] The determination unit 83 is further configured to determine that the second target road line is an outlier line when an angle between a main direction of the second target road line and a main direction of the target point cluster is greater than or equal to an angle threshold.

[0188] In some embodiments, the point cloud included in each of the N point clusters is also used to indicate at least one stop line, and the determination unit 83 is also used to determine a stop line indicated by the target point cluster as a first target stop line, and determine a second inlier rate of the point cloud corresponding to the first target stop line based on an inlier rate algorithm corresponding to the stop line; when the second inlier rate is less than or equal to an inlier rate threshold corresponding to the stop line, the first target stop line is determined to be a deformed line.

[0189] In some embodiments, the determination unit 83 is also used to determine a stop line indicated by the target point cluster as a second target stop line, determine the center of mass of the point cloud corresponding to the second target stop line; determine the second distance between the center of mass of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster; when the second distance is greater than or equal to a second distance threshold, determine the second target stop line as an outlier line.

[0190] In some embodiments, the singular value decomposition unit 82 is further used to perform singular value decomposition on the point cloud corresponding to the second target stop line when the second distance is less than the second distance threshold, to obtain a second singular vector, and the second singular vector is used to indicate the main direction of the second target stop line;

[0191] The determining unit 83 is further configured to determine that the second target stop line is an outlier line when the angle between the main direction of the second target stop line and the main direction of the target point cluster is not within a preset angle range.

[0192] It should be noted that the above-mentioned road data analysis device can be the electronic device in the above-mentioned method embodiment of the present application, or it can be a functional module and / or functional entity in the electronic device that can realize the function of the device embodiment, and the embodiment of the present application is not limited.

[0193] In the embodiments of the present application, each module can implement the road data analysis method provided in the above method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0194] Fig. 9 The structural diagram of an electronic device provided in the embodiment of the present application is used to exemplify the electronic device for implementing any road data analysis method in the embodiment of the present application, and should not be understood as a specific limitation on the embodiment of the present application.

[0195] like Fig. 9 As shown, the electronic device 900 may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage device 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device 900 are also stored. The processor 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0196] Typically, the following devices may be connected to the I / O interface 905: input devices 906 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 907 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 908 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 909. The communication device 909 may allow the electronic device 900 to communicate wirelessly or wired with other devices to exchange data. Although an electronic device 900 having various devices is shown, it should be understood that it is not required to implement or have all the devices shown. More or fewer devices may be implemented or have alternatively.

[0197] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 909, or installed from the storage device 908, or installed from the ROM 902. When the computer program is executed by the processor 901, the functions defined in any road data analysis method provided in the embodiment of the present application can be executed.

[0198] It should be noted that the computer-readable medium mentioned above in the present application may 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 be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, 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, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0199] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0200] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0201] The computer-readable medium carries one or more programs, and when the one or more programs are executed by the electronic device, the electronic device implements any road data analysis method.

[0202] In an embodiment of the present application, a computer program code for performing the operation of the present application can be written in one or more programming languages ​​or a combination thereof, and the above-mentioned programming languages ​​include but are not limited to object-oriented programming languages, such as Java, Smalltalk, C++, and also include conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed completely on the computer, partially on the computer, as an independent software package, partially on the computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0203] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0204] The units involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of a unit does not, in some cases, constitute a limitation on the unit itself.

[0205] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0206] In the context of the present application, a computer-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a computer-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, 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 foregoing.

[0207] An embodiment of the present application also provides a vehicle, comprising: the above-mentioned road data analysis device, or the above-mentioned electronic device, or the above-mentioned computer-readable storage medium.

[0208] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

[0209] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0210] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. A road data analysis method, It is characterized in that The method comprises: Clustering the road point cloud in the BEV road perception image according to the point spacing to obtain N point clusters, where N is a positive integer, each point cluster in the N point clusters is used to indicate a different road segment, and the point cloud included in each point cluster in the N point clusters is used to indicate at least one target road line, where the target road line is a lane line or a road edge line; Taking one of the N point clusters as a target point cluster, performing singular value decomposition on the target point cluster to obtain a first singular value and a second singular value, wherein the first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster; Determining a road type of a target road segment indicated by the target point cluster based on a difference between the first singular value and the second singular value; Determine a target road line indicated by the target point cluster as a first target road line, and determine a first inlier rate of a point cloud corresponding to the first target road line based on an inlier rate algorithm corresponding to the road type; When the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line.

2. The method according to claim 1, It is characterized in that The determining, based on the difference between the first singular value and the second singular value, the road type of the target road segment indicated by the target point cluster comprises: Determine a first parameter according to the first singular value and the second singular value, where the first parameter is used to indicate a difference between the first singular value and the second singular value; When the first parameter is greater than or equal to a first threshold, determining that the road type of the target road segment indicated by the target point cluster is a straight road; When the first parameter is less than the first threshold value and greater than the second threshold value, determining that the road type of the target road segment indicated by the target point cluster is a small curvature curve; When the first parameter is less than or equal to the second threshold, determining that the road type of the target road segment indicated by the target point cluster is a large curvature curve; The first threshold is greater than the second threshold, and the curvature of the small curvature curve is smaller than the curvature of the large curvature curve.

3. The method according to claim 2, It is characterized in that The step of determining a first parameter according to the first singular value and the second singular value comprises: Determine a ratio of the first singular value to the second singular value as the first parameter; Or, a difference between the first singular value and the second singular value is determined as the first parameter.

4. The method according to claim 1, It is characterized in that The determining, based on the inlier rate algorithm corresponding to the road type, a first inlier rate of the point cloud corresponding to the first target road line comprises: Determining a corresponding preset number according to the road type; generating a reference line corresponding to the road type based on the preset number of reference points randomly selected from the point cloud of the first target road line; Determine the number of points that meet a target condition from other points in the point cloud of the first target road line, wherein the target condition is that the distance from the point to the reference line is less than or equal to a reference threshold corresponding to the road type; The first inlier rate is determined based on a ratio of the number of points satisfying the target condition to the total number of point clouds corresponding to the first target road line.

5. The method according to claim 1, It is characterized in that The method further comprises: Determine a target road line indicated by the target point cluster as a second target road line, and determine the centroid of the point cloud corresponding to the second target road line; Determine a first distance between the centroid of the point cloud corresponding to the second target road line and the cluster center of the target point cluster; When the first distance is greater than or equal to a first distance threshold, determining that the second target road line is an outlier line; When the first distance is less than the first distance threshold, performing singular value decomposition on the second target road line to obtain a first singular vector, where the first singular vector is used to indicate a main direction of the second target road line; When the angle between the main direction of the second target road line and the main direction of the target point cluster is greater than or equal to an angle threshold, the second target road line is determined to be an outlier line.

6. The method according to claim 1, It is characterized in that The point cloud included in each of the N point clusters is also used to indicate at least one stop line, and the method further includes: Determine a stop line indicated by the target point cluster as a first target stop line, and determine a second inlier rate of the point cloud corresponding to the first target stop line based on an inlier rate algorithm corresponding to the stop line; When the second inlier rate is less than or equal to the inlier rate threshold corresponding to the stop line, the first target stop line is determined to be a deformed line.

7. The method according to claim 6, It is characterized in that The method further comprises: Determine a stop line indicated by the target point cluster as a second target stop line, and determine the centroid of the point cloud corresponding to the second target stop line; Determine a second distance between the centroid of the point cloud corresponding to the second target stop line and the cluster center of the target point cluster; When the second distance is greater than or equal to a second distance threshold, determining that the second target stop line is an outlier line; When the second distance is less than the second distance threshold, performing singular value decomposition on the point cloud corresponding to the second target stop line to obtain a second singular vector, where the second singular vector is used to indicate a main direction of the second target stop line; When the angle between the main direction of the second target stop line and the main direction of the target point cluster is not within a preset angle range, the second target stop line is determined to be an outlier line.

8. A road data analysis device, It is characterized in that include: A clustering unit, used for clustering the road point cloud in the BEV road perception image according to the point spacing to obtain N point clusters, where N is a positive integer, each of the N point clusters is used to indicate a different road segment, and the point cloud included in each of the N point clusters is used to indicate at least one target road line, where the target road line is a lane line or a road edge line; A singular value decomposition unit, used to take one of the N point clusters as a target point cluster, perform singular value decomposition on the target point cluster, and obtain a first singular value and a second singular value, wherein the first singular value is used to indicate the variance of the point cloud distribution of the target point cluster in the main direction of the target point cluster, and the second singular value is used to indicate the variance of the point cloud distribution of the target point cluster in a direction perpendicular to the main direction; the main direction of the target point cluster is the driving direction of the road segment indicated by the target point cluster; a determining unit, configured to determine a road type of a target road segment indicated by the target point cluster based on a difference between the first singular value and the second singular value; Determine a target road line indicated by the target point cluster as a first target road line, and determine a first inlier rate of a point cloud corresponding to the first target road line based on an inlier rate algorithm corresponding to the road type; When the first inlier rate is less than or equal to an inlier rate threshold corresponding to the road type, the first target road line is determined to be a deformed line.

9. An electronic device, It is characterized in that include: A processor, wherein the processor is used to execute a computer program stored in a memory, wherein the computer program, when executed by the processor, implements the steps of the road data analysis method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the road data analysis method according to any one of claims 1 to 7 are implemented.

11. A vehicle, It is characterized in that include: The road data analysis device as claimed in claim 8, or the electronic device as claimed in claim 9, or includes the computer-readable storage medium as claimed in claim 10.

Images