Map road edge generation method, device, system and readable storage medium

Abstract

The application discloses a kind of map road edge line generation method, device, system and readable storage medium, the method includes: obtaining the point cloud data of target road, and according to the point cloud data determines point cluster set;Determine the target straight line model corresponding to each point cluster in the point cluster set, obtain target straight line model set;According to the target straight line model set generates the map road edge line corresponding to the target road.The application determines the point cluster set of point cloud data, determines the target straight line model of each point cluster in the point cluster set, and then constructs the map road edge line corresponding to the target road by target straight line model, so that it is not necessary to rely on artificial to make the map road edge line of point cloud data, improves the generation efficiency of map road edge line.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of map generation, in particular to a map road edge generation method, device, system and readable storage medium. BACKGROUND

[0002] Compared with high-speed and urban outdoor scenes, vehicles lack GPS information as a global positioning reference in indoor scenes (such as parking lots, etc.), and there are sparse positioning elements in some areas of indoor scenes such as long straight roads and spiral passages. At this time, vehicles mainly rely on perception modules to obtain data matching with road edges in indoor maps to complete real-time position correction. Therefore, road edges are an important element indispensable to the indoor map generation link.

[0003] Due to objective factors such as sensor measurement noise, positioning cumulative error and trajectory alignment error, the point cloud data participating in indoor map mapping has a serious divergence problem. Traditional algorithms cannot handle seriously divergent point cloud data and can only rely on manual production of road edges of indoor maps according to point cloud data, resulting in low efficiency of generating map road edges.

[0004] Therefore, how to improve the generation efficiency of map road edges is a problem that needs to be solved. SUMMARY

[0005] The main purpose of the present application is to provide a map road edge generation method, device, system and readable storage medium, which aims to improve the generation efficiency of map road edges.

[0006] To achieve the above purpose, the present application provides a map road edge generation method, which comprises the following steps:

[0007] Obtain point cloud data of a target road, and determine a point cluster set according to the point cloud data;

[0008] Determine a target straight line model corresponding to each point cluster in the point cluster set to obtain a target straight line model set;

[0009] Generate a map road edge corresponding to the target road according to the target straight line model set.

[0010] Optionally, the step of determining a point cluster set according to the point cloud data comprises:

[0011] Downsample the point cloud data, and cluster the downsampled point cloud data to segment the downsampled point cloud data into a plurality of spatially discontinuous point clusters to obtain the point cluster set.

[0012] Optionally, the step of determining a target straight line model corresponding to each point cluster in the point cluster set to obtain a target straight line model set comprises:

[0013] selecting a point cluster from the point cluster set in sequence, and determining a line model set corresponding to the selected point cluster in a manner of determining a line model with two points in the selected point cluster;

[0014] determining an inlier number and a symmetry segmentation ability corresponding to each line model in the line model set according to a first preset threshold value;

[0015] determining a target line model corresponding to the selected point cluster from the line model set based on the inlier number and the symmetry segmentation ability;

[0016] until a target line model corresponding to each point cluster in the point cluster set is determined, a target line model set corresponding to the point cluster set is obtained.

[0017] Optionally, the step of determining a target line model corresponding to each point cluster in the point cluster set to obtain a target line model set comprises:

[0018] selecting a point cluster from the point cluster set in sequence, and randomly determining two points in the selected point cluster to determine a first line model corresponding to the selected point cluster;

[0019] updating the first line model according to a first preset threshold value and a distance between each point in the selected point cluster and the first line model to obtain a target line model corresponding to the selected point cluster;

[0020] until a target line model corresponding to each point cluster in the point cluster set is determined, a target line model set corresponding to the point cluster set is obtained.

[0021] Optionally, the first preset threshold value comprises an inlier threshold value and a divergence threshold value, and the step of updating the first line model according to a first preset threshold value and a distance between each point in the selected point cluster and the first line model to obtain a target line model corresponding to the selected point cluster comprises:

[0022] determining an inlier rate corresponding to the first line model according to the distance between each point in the selected point cluster and the first line model and the inlier threshold value;

[0023] calculating an update iteration number according to the inlier rate and a confidence degree corresponding to the first line model;

[0024] updating the first line model based on the update iteration number, the inlier threshold value and the divergence threshold value to obtain a target line model corresponding to the selected point cluster.

[0025] Optionally, the step of updating the first straight line model based on the update iteration number, the inlier threshold and the divergence threshold to obtain a target straight line model corresponding to the selected point cluster comprises:

[0026] randomly determining two points in the selected point cluster, and determining a second straight line model corresponding to the selected point cluster according to the two selected points;

[0027] calculating a distance between each point in the selected point cluster and the first straight line model and a distance between each point in the selected point cluster and the second straight line model, respectively;

[0028] determining an inlier number of the first straight line model according to the distance between each point in the selected point cluster and the first straight line model and the inlier threshold, and determining an inlier number of the second straight line model according to the distance between each point in the selected point cluster and the second straight line model and the inlier threshold;

[0029] comparing the inlier number of the second straight line model with the inlier number of the first straight line model;

[0030] if the inlier number of the second straight line model is greater than the inlier number of the first straight line model, taking the second straight line model as the first straight line model to update the first straight line model;

[0031] if the inlier number of the second straight line model is not greater than the inlier number of the first straight line model, determining a symmetric segmentation ability of the first straight line model and the second straight line model according to the inlier threshold and the divergence threshold, and updating the first straight line model according to the symmetric segmentation ability of the first straight line model and the second straight line model;

[0032] determining a target straight line model corresponding to the selected point cluster based on an update result of updating the first straight line model.

[0033] Optionally, the step of determining a target straight line model corresponding to the selected point cluster based on an update result of updating the first straight line model comprises:

[0034] determining a current cumulative iteration number and an inlier rate of a current first straight line model;

[0035] comparing the cumulative iteration number with the update iteration number, and comparing the inlier rate of the current first straight line model with a second preset threshold;

[0036] if the cumulative iteration number is greater than or equal to the update iteration number, or the inlier rate is greater than the second preset threshold, taking the current first straight line model as a target straight line model corresponding to the selected point cluster;

[0037] If the accumulated iteration number is less than the updated iteration number, and the inlier ratio is not greater than the second preset threshold, re-performing the step of randomly determining two points in the selected point cluster to determine a second straight line model corresponding to the selected point cluster.

[0038] Optionally, the step of determining the symmetric partitioning ability of the first straight line model and the second straight line model according to the inlier threshold and the divergent threshold comprises:

[0039] determining a first accumulated distance of all points in the selected point cluster having a distance less than a preset multiple of the inlier threshold to the first straight line model to the first straight line model;

[0040] determining a second accumulated distance of all points in the selected point cluster having a distance greater than the divergent threshold to the first straight line model to the first straight line model;

[0041] determining the symmetric partitioning ability of the first straight line model based on the first accumulated distance and the second accumulated distance;

[0042] determining a third accumulated distance of all points in the selected point cluster having a distance less than a preset multiple of the inlier threshold to the second straight line model to the second straight line model;

[0043] determining a fourth accumulated distance of all points in the selected point cluster having a distance greater than the divergent threshold to the second straight line model to the second straight line model;

[0044] determining the symmetric partitioning ability of the second straight line model based on the third accumulated distance and the fourth accumulated distance.

[0045] Optionally, before the step of updating the first straight line model according to the symmetric partitioning ability of the first straight line model and the second straight line model, further comprising:

[0046] comparing the symmetric partitioning ability of the first straight line model and the symmetric partitioning ability of the second straight line model;

[0047] if the symmetric partitioning ability of the first straight line model is superior to the symmetric partitioning ability of the second straight line model, not updating the first straight line model, and performing the step of determining a current accumulated iteration number and an inlier ratio of a current first straight line model;

[0048] the step of updating the first straight line model according to the symmetric partitioning ability of the first straight line model and the second straight line model comprises:

[0049] if the symmetric partitioning ability of the second straight line model is superior to the symmetric partitioning ability of the first straight line model, taking the second straight line model as the first straight line model to update the first straight line model.

[0050] Optionally, the step of generating the map road edge line corresponding to the target road according to the target straight line model set comprises:

[0051] determining an outlier corresponding to each target straight line model in the target straight line model set;

[0052] comparing the number of outliers with a third preset threshold value;

[0053] if the number of outliers is less than the third preset threshold value, generating the map road edge line corresponding to the target road according to the target straight line model set;

[0054] if the number of outliers is not less than the third preset threshold value, constructing new point cloud data based on the outliers, and re-executing the step of determining the point cluster set according to the point cloud data based on the new point cloud data.

[0055] In addition, to achieve the above object, the application further provides a map road edge line generation device, which comprises:

[0056] a first determination module configured to acquire point cloud data of a target road and determine a point cluster set according to the point cloud data;

[0057] a second determination module configured to determine a target straight line model corresponding to each point cluster in the point cluster set, thereby obtaining a target straight line model set;

[0058] a generation module configured to generate a map road edge line corresponding to the target road according to the target straight line model set.

[0059] Further, the first determination module is further configured to:

[0060] down-sample the point cloud data, cluster the down-sampled point cloud data, and divide the down-sampled point cloud data into a plurality of spatially discontinuous point clusters, thereby obtaining a point cluster set.

[0061] Further, the second determination module is further configured to:

[0062] select a point cluster from the point cluster set in sequence, and determine a straight line model set corresponding to the selected point cluster in the selected point cluster by determining a straight line model with two points;

[0063] determine the number of inliers and the symmetric segmentation capability corresponding to each straight line model in the straight line model set according to a first preset threshold value;

[0064] determine a target straight line model corresponding to the selected point cluster from the straight line model set based on the number of inliers and the symmetric segmentation capability.

[0065] until determining the target straight line model corresponding to each point cluster in the point cluster set, obtaining the target straight line model set corresponding to the point cluster set.

[0066] Further, the second determination module is further used for:

[0067] selecting a point cluster in the point cluster set in sequence, and randomly determining two points in the selected point cluster, to determine a first straight line model corresponding to the selected point cluster;

[0068] updating the first straight line model according to the first preset threshold and the distance between each point in the selected point cluster and the first straight line model, to obtain a target straight line model corresponding to the selected point cluster;

[0069] until determining the target straight line model corresponding to each point cluster in the point cluster set, obtaining the target straight line model set corresponding to the point cluster set.

[0070] Further, the second determination module is further used for:

[0071] determining an inlier rate corresponding to the first straight line model according to the inlier threshold and the distance between each point in the selected point cluster and the first straight line model;

[0072] calculating an update iteration number according to the confidence corresponding to the first straight line model and the inlier rate;

[0073] updating the first straight line model based on the update iteration number, the inlier threshold and the divergence threshold, to obtain a target straight line model corresponding to the selected point cluster.

[0074] Further, the second determination module is further used for:

[0075] randomly determining two points in the selected point cluster, and determining a second straight line model corresponding to the selected point cluster according to the selected two points;

[0076] respectively calculating the distance between each point in the selected point cluster and the first straight line model and the distance between each point in the selected point cluster and the second straight line model;

[0077] determining the inlier number of the first straight line model according to the distance between each point in the selected point cluster and the first straight line model and the inlier threshold, and determining the inlier number of the second straight line model according to the distance between each point in the selected point cluster and the second straight line model and the inlier threshold;

[0078] comparing the inlier number of the second straight line model with the inlier number of the first straight line model;

[0079] If the number of interior points of the second straight line model is greater than the number of interior points of the first straight line model, then the second straight line model is used as the first straight line model to update the first straight line model.

[0080] If the number of interior points of the second straight line model is not greater than the number of interior points of the first straight line model, then the symmetric segmentation capability of the first straight line model and the second straight line model is determined according to the interior point threshold and the divergence threshold, and the first straight line model is updated according to the symmetric segmentation capability of the first straight line model and the second straight line model.

[0081] Based on the update result of the first straight line model, the target straight line model corresponding to the selected point cluster is determined.

[0082] Furthermore, the second determining module is also used for:

[0083] Determine the current cumulative iteration count and the interior point rate of the current first straight line model;

[0084] The cumulative number of iterations is compared with the number of update iterations, and the inlier rate of the current first straight line model is compared with the second preset threshold.

[0085] If the cumulative number of iterations is greater than or equal to the number of update iterations, or the inlier rate is greater than the second preset threshold, then the current first straight line model is used as the target straight line model corresponding to the selected point cluster.

[0086] If the cumulative iteration count is less than the update iteration count, and the interior point rate is not greater than the second preset threshold, then the following step is executed again: randomly determine two points in the selected point cluster to determine the second straight line model corresponding to the selected point cluster.

[0087] Furthermore, the second determining module is also used for:

[0088] Determine the first cumulative distance from the selected point cluster to the first straight line model for all points whose distance from the first straight line model is less than a preset multiple of the in-point threshold.

[0089] Determine the second cumulative distance from the selected point cluster to the first straight line model for all points whose distance from the first straight line model is greater than the divergence threshold;

[0090] The symmetric segmentation capability of the first straight line model is determined based on the first cumulative distance and the second cumulative distance;

[0091] The third cumulative distance from the selected point cluster to the second straight line model is determined for all points whose distance from the second straight line model is less than the preset multiple of the interior point threshold.

[0092] determine a fourth accumulated distance of all points in the selected point cluster that are greater than the divergence threshold from the second straight line model to the second straight line model;

[0093] determine the symmetry segmentation capability of the second straight line model based on the third accumulated distance and the fourth accumulated distance.

[0094] Further, the second determining module is further configured to:

[0095] compare the symmetry segmentation capability of the first straight line model with the symmetry segmentation capability of the second straight line model;

[0096] if the symmetry segmentation capability of the first straight line model is better than the symmetry segmentation capability of the second straight line model, then the first straight line model is not updated, and a step of determining a current accumulated iteration number and a current in-point rate of the first straight line model is performed;

[0097] if the symmetry segmentation capability of the second straight line model is better than the symmetry segmentation capability of the first straight line model, then the second straight line model is taken as the first straight line model to update the first straight line model.

[0098] Further, the generating module is further configured to:

[0099] determine an out-point corresponding to each target straight line model in the target straight line model set;

[0100] compare the number of out-points with a third preset threshold;

[0101] if the number of out-points is less than the third preset threshold, then a map road edge line corresponding to the target road is generated according to the target straight line model set;

[0102] if the number of out-points is not less than the third preset threshold, then new point cloud data is constituted based on the out-points, and the step of determining a point cluster set according to the point cloud data is re-executed based on the new point cloud data.

[0103] In addition, to achieve the above object, the present application further provides a map road edge line generation system, which comprises a memory, a processor and a map road edge line generation program stored in the memory and executable on the processor, and the map road edge line generation program implements the steps of the above map road edge line generation method when executed by the processor.

[0104] In addition, to achieve the above object, the present application further provides a readable storage medium, which stores a map road edge line generation program, and the map road edge line generation program implements the steps of the above map road edge line generation method when executed by a processor.

[0105] The map road edge generation method provided by the application obtains point cloud data of a target road, determines a point cluster set according to the point cloud data, determines a target straight line model corresponding to each point cluster in the point cluster set to obtain a target straight line model set, and generates a map road edge corresponding to the target road according to the target straight line model set. The point cluster set of the point cloud data is determined, the target straight line model of each point cluster in the point cluster set is determined, and then the map road edge corresponding to the target road is constructed through the target straight line model, so that the map road edge is not required to be manually made according to the point cloud data, and the generation efficiency of the map road edge is improved. BRIEF DESCRIPTION OF DRAWINGS

[0106] Figure 1 is a device structure schematic diagram of a hardware running environment related to an embodiment scheme of the application.

[0107] Figure 2 is a flowchart of a first embodiment of the map road edge generation method of the application.

[0108] Figure 3 is a flowchart of a second embodiment of the map road edge generation method of the application.

[0109] Figure 4 is a flowchart of a third embodiment of the map road edge generation method of the application.

[0110] Figure 5 is a schematic diagram of the relationship between a straight line model and an inner point threshold value and a divergence threshold value.

[0111] Figure 6 is a flowchart of a fourth embodiment of the map road edge generation method of the application.

[0112] Figure 7 is a flowchart of a fifth embodiment of the map road edge generation method of the application.

[0113] Figure 8 is a device structure schematic diagram of the map road edge generation apparatus.

[0114] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0115] It should be understood that the specific embodiments described herein are merely intended to explain the application and are not intended to limit the application.

[0116] As Figure 1 shown, Figure 1 is a device structure schematic diagram of a hardware running environment related to an embodiment scheme of the application.

[0117] The device of the embodiment of the application can be a PC or a server device.

[0118] As shown in Figure 1 , the device can include a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between the components. The user interface 1003 can include a display, an input unit such as a keyboard, and can also include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface). The memory 1005 can be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 can also be a storage device independent of the aforementioned processor 1001.

[0119] Those skilled in the art can understand that Figure 1 the device structure shown in the above description does not constitute a limitation on the device, and can include more or fewer components than shown, or combine certain components, or different component arrangements.

[0120] As shown in Figure 1 , the memory 1005, as a computer storage medium, can include an operating system, a network communication module, a user interface module, and a map road edge line generation program.

[0121] The operating system is a program that manages and controls the portable map road edge line generation system and software resources, supports the running of the network communication module, the user interface module, the map road edge line generation program, and other programs or software; the network communication module is used to manage and control the network interface 1002; and the user interface module is used to manage and control the user interface 1003.

[0122] In the map road edge line generation system shown in Figure 1 , the map road edge line generation system calls the map road edge line generation program stored in the memory 1005 through the processor 1001, and performs the operations in each embodiment of the map road edge line generation method described below.

[0123] Based on the above hardware structure, embodiments of the map road edge line generation method of the application are proposed.

[0124] Referring to Figure 2 , Figure 2A flowchart of a first embodiment of a map road edge generation method of the present application is shown. The map road edge generation method of the present embodiment is applied to a map generation system, which can be applied to intelligent terminals, PC terminals and other devices. The method comprises:

[0125] In step S10, point cloud data of a target road is obtained, and a point cluster set is determined according to the point cloud data.

[0126] In the present embodiment, the map generation system obtains point cloud data of a target road, and performs downsampling and clustering on the point cloud data, thereby obtaining a point cluster set corresponding to the point cloud data. It can be understood that when the point cloud data of the target road is needed, relevant personnel drives a vehicle equipped with a corresponding sensor on the target road, thereby collecting the point cloud data of the target road through the sensor, and sending the point cloud data to the map generation system. The point cloud data can be ultrasonic point cloud data, lane point cloud data, laser point cloud data, trajectory point cloud data, etc. Preferably, the point cloud data is ultrasonic point cloud data, and the corresponding sensor is an ultrasonic sensor.

[0127] Specifically, the step of determining a point cluster set according to the point cloud data comprises:

[0128] In step S101, the point cloud data is downsampled, and the downsampled point cloud data is clustered to obtain a point cluster set.

[0129] In this step, after obtaining the point cloud data of the target road, the map generation system performs downsampling on the point cloud data, and clusters the downsampled point cloud data to automatically segment the point cloud data into a plurality of spatially discontinuous point clusters, thereby obtaining a point cluster set corresponding to the point cloud data.

[0130] It should be noted that the point cloud data is downsampled by a point cloud filtering method, which includes straight-through filtering, voxel filtering, statistical filtering, conditional filtering, statistical filtering, etc. Preferably, the point cloud data is downsampled by voxel filtering, which can achieve downsampling while not destroying the geometric structure of the point cloud data itself, thereby not only reducing the time-consuming of subsequent calculation, but also ensuring the accuracy of subsequent calculation. The downsampled point cloud data is clustered by a density clustering algorithm, which includes DBSCAN, OPTICS, DENCLUE, etc. Preferably, DBSCAN is used. DBSCAN defines a point cluster as the maximum set of points connected in density, which can divide all point clouds in a region with sufficiently high density into the same point cluster. By dividing the point cluster, the time-consuming of calculation is reduced, and the accuracy of determining the straight line model is improved.

[0131] Step S20, determine the target straight line model corresponding to each point cluster in the point cluster set, and obtain a target straight line model set. In this embodiment, after the map generation system determines the point cluster set corresponding to the point cloud data, the target straight line model of each point cluster in the point cluster set is determined by using an improved RANSAC (Random Sample Consensus) algorithm, and the target straight line model can express the real road edge corresponding to the point cluster.

[0132] Specifically, step S20 includes:

[0133] Step S201, sequentially select a point cluster from the point cluster set, and randomly determine two points in the selected point cluster to determine a first straight line model corresponding to the selected point cluster.

[0134] In this step, the map generation system sequentially selects a point cluster from the point cluster set, and randomly determines two points in the selected point cluster to determine a first straight line model corresponding to the selected point cluster. It can be understood that the point cluster includes a plurality of points, and the density of the points in the point cluster gradually decreases from the center to the edge, that is, the density of the points in the center of the point cluster is relatively large, and the density of the points at the edge of the point cluster is relatively small. Preferably, the map generation system randomly selects two points in the area with a relatively large density of points in the point cluster to determine the first straight line model.

[0135] Step S202, update the first straight line model according to the first preset threshold and the distance between each point in the selected point cluster and the first straight line model, to obtain a target straight line model corresponding to the selected point cluster.

[0136] In this step, the map generation system evaluates the first straight line model according to the first preset threshold and the distance between each point in the selected point cluster and the first straight line model, and updates the first straight line model according to the evaluation result and a preset number of update iterations, to obtain a target straight line model corresponding to the selected point cluster.

[0137] Step S203, until the target straight line model corresponding to each point cluster in the point cluster set is determined, a target straight line model set corresponding to the point cluster set is obtained.

[0138] In this step, the map generation system sequentially executes the above steps S201 and S202 on each point cluster in the point cluster set, until the target straight line model corresponding to each point cluster in the point cluster set is determined, and a target straight line model set corresponding to the point cluster set is obtained.

[0139] Step S30, generate a map road edge corresponding to the target road according to the target straight line model set.

[0140] In this embodiment, after determining the target straight line model set corresponding to the point cluster set, the map generation system connects each target straight line model in the target straight line model set according to the geometric structure of the point cloud data itself to generate the map road edge line corresponding to the target road.

[0141] The map generation system of this embodiment acquires point cloud data of a target road and determines a set of point clusters based on the point cloud data; it then determines the target straight line model corresponding to each point cluster in the point cluster set, obtaining a set of target straight line models; and finally, it generates the map road edge line corresponding to the target road based on the set of target straight line models. This invention improves the efficiency of map road edge line generation by determining the point cluster set of point cloud data, determining the target straight line model for each point cluster in the point cluster set, and then constructing the map road edge line corresponding to the target road using the target straight line model. This eliminates the need for manual creation of map road edge lines based on point cloud data.

[0142] Further, refer to Figure 3 Based on the first embodiment of the map road edge line generation method of the present invention, a second embodiment of the map road edge line generation method of the present invention is proposed.

[0143] The difference between the second embodiment of the map road edge generation method and the first embodiment of the map road edge generation method is that step S202 in the first embodiment includes the following steps:

[0144] Step S2021: Determine the inlier rate corresponding to the first straight line model based on the inlier threshold and the distance from each point in the selected point cluster to the first straight line model.

[0145] In this embodiment, the first preset threshold includes an interior point threshold, which is determined based on sensor measurement error. For ultrasonic point cloud data, the corresponding sensor is an ultrasonic sensor. The map generation system determines the interior point ratio of the selected point cluster under the first straight line model based on the interior point threshold and the distance from each point in the selected point cluster to the first straight line model. Specifically, the map generation system compares the distance from each point in the selected point cluster to the first straight line model with the interior point threshold to determine that points whose distance to the first straight line model is less than the interior point threshold are interior points of the first straight line model. Based on the number of interior points of the first straight line model and the total number of points contained in the selected point cluster, the system determines the interior point ratio corresponding to the first straight line model.

[0146] Step S2022: Calculate the number of update iterations based on the confidence level and the inlier rate corresponding to the first straight line model;

[0147] In this embodiment, the map generation system calculates the number of update iterations of the first linear model corresponding to the selected point cluster according to the inlier rate of the first linear model and the confidence degree corresponding to the first linear model. It should be noted that the confidence degree of the first linear model is artificially specified, and the confidence degree refers to the probability that two points randomly selected from the selected point cluster are inliers of the first linear model. Specifically, the formula for calculating the number of update iterations is as follows:

[0148] represents the upward rounding.

[0149] wherein n is the number of update iterations, q is the confidence degree of the first linear model, and p is the inlier rate.

[0150] In step S2023, the first linear model is updated based on the number of update iterations, the inlier threshold, and the divergence threshold to obtain a target linear model corresponding to the selected point cluster.

[0151] In this embodiment, the first preset threshold includes an inlier threshold and a divergence threshold. The map generation system updates the first linear model based on the number of update iterations, the inlier threshold, and the divergence threshold to obtain a target linear model corresponding to the selected point cluster. It should be noted that the divergence threshold is determined according to the divergence degree of the point cloud data. The maximum width of the point cloud data is obtained, the divergence threshold is determined according to the maximum width, and all point clusters in the point cluster set share the same divergence threshold.

[0152] Specifically, step S2023 includes:

[0153] In step S20231, two points are randomly determined in the selected point cluster, and a second linear model corresponding to the selected point cluster is determined according to the two selected points.

[0154] In this step, the map generation system randomly determines two points in the selected point cluster, and determines the first linear model of the selected point cluster based on the two selected points. It can be understood that the point cluster includes a plurality of points, and the density of the points in the point cluster gradually decreases from the center to the edge, that is, the density of the points in the center of the point cluster is larger, and the density of the points at the edge of the point cluster is smaller. Preferably, the map generation system randomly selects two points in the area with a larger density of points in the point cluster to determine the second linear model, wherein the second linear model is different from the first linear model.

[0155] In step S201232, the distance of each point in the selected point cluster to the first linear model and the distance of each point in the selected point cluster to the second linear model are calculated respectively.

[0156] In this step, the map generation system calculates the distance of each point in the selected point cluster to the first straight line model and the distance of each point in the selected point cluster to the second straight line model respectively; specifically, the distance of each point to the first straight line model and the distance of each point to the second straight line model can be determined according to the coordinate value of each point and the coordinate value of the first straight line model (or the second straight line model). The specific distance calculation method is not limited here.

[0157] In step S20233, the number of inliers of the first straight line model is determined according to the distance of each point in the selected point cluster to the first straight line model and the inlier threshold, and the number of inliers of the second straight line model is determined according to the distance of each point in the selected point cluster to the second straight line model and the inlier threshold.

[0158] In this step, the map generation system compares the distance of each point in the selected point cluster to the first straight line model with the inlier threshold, determines that the point whose distance to the first straight line model is less than the inlier threshold is an inlier of the first straight line model, and counts the number of inliers of the first straight line model; the map generation system compares the distance of each point in the selected point cluster to the second straight line model with the inlier threshold, determines that the point whose distance to the second straight line model is less than the inlier threshold is an inlier of the second straight line model, and counts the number of inliers of the second straight line model; it can be understood that since the inlier threshold is determined according to the sensor measurement error, the inlier thresholds corresponding to the first straight line model and the second straight line model are the same.

[0159] In step S20234, the number of inliers of the second straight line model is compared with the number of inliers of the first straight line model.

[0160] In this step, the map generation system compares the number of inliers of the second straight line model with the number of inliers of the first straight line model.

[0161] In step S20235, if the number of inliers of the second straight line model is greater than the number of inliers of the first straight line model, the second straight line model is taken as the first straight line model to update the first straight line model.

[0162] In this step, after the map generation system compares the number of inliers of the second straight line model with the number of inliers of the first straight line model, if the comparison result is that the number of inliers of the second straight line model is greater than the number of inliers of the first straight line model, it is determined that the second straight line model is better than the first straight line model, and the second straight line model is taken as the first straight line model, i.e., the first straight line model is updated.

[0163] In step S20236, if the number of inliers of the second straight line model is not greater than the number of inliers of the first straight line model, the symmetry segmentation capability of the first straight line model and the second straight line model is determined according to the inlier threshold and the divergence threshold, and the first straight line model is updated according to the symmetry segmentation capability of the first straight line model and the second straight line model.

[0164] In this step, after the map generation system compares the number of inliers of the second straight line model with the number of inliers of the first straight line model, if the comparison result is that the number of inliers of the second straight line model is not greater than the number of inliers of the first straight line model, the symmetry segmentation capability of the first straight line model and the second straight line model is determined according to the inlier threshold and the divergence threshold, in combination with the distance of each point in the selected point cluster to the first straight line model and the second straight line model, and the symmetry segmentation capability of the first straight line model and the second straight line model is compared, if the symmetry segmentation capability of the first straight line model is better than that of the second straight line model, the first straight line model is not updated, that is, the first straight line model is not replaced by the second straight line model, if the symmetry segmentation capability of the second straight line model is better than that of the first straight line model, the first straight line model is updated, that is, the second straight line model is taken as the first straight line model.

[0165] In step S20237, the target straight line model corresponding to the selected point cluster is determined based on the update result of the updated first straight line model.

[0166] Further, step S20237 includes:

[0167] In step S202371, the current cumulative iteration number and the inlier rate of the current first straight line model are determined.

[0168] In this step, the map generation system adds 1 to the current cumulative iteration number each time the first straight line model is updated, and then determines the inlier rate of the selected point cluster under the current first straight line model according to the inlier threshold and the distance of each point in the selected point cluster to the current first straight line model.

[0169] In step S202372, the cumulative iteration number is compared with the update iteration number, and the inlier rate of the current first straight line model is compared with a second preset threshold.

[0170] In this step, the map generation system compares the cumulative iteration number with the update iteration number, and compares the inlier rate of the current first straight line model with a second preset threshold; the second preset threshold is set in advance, such as 50%, 60%, etc., which is not limited here.

[0171] Step S202373: If the cumulative number of iterations is greater than or equal to the number of update iterations, or the inlier rate is greater than the second preset threshold, then the current first straight line model is used as the target straight line model corresponding to the selected point cluster.

[0172] In this step, the map generation system compares the cumulative number of iterations with the number of update iterations, and compares the inlier rate of the current first straight line model with the second preset threshold. If the comparison result is that the cumulative number of iterations is greater than or equal to the number of update iterations, or the inlier rate is greater than the second preset threshold, the iteration loop is exited, and the current first straight line model is used as the target straight line model corresponding to the selected point cluster.

[0173] Step S202374: If the cumulative iteration count is less than the update iteration count, and the interior point rate is not greater than the second preset threshold, then the following step is executed again: randomly determine two points in the selected point cluster to determine the second straight line model corresponding to the selected point cluster.

[0174] In this step, the map generation system compares the cumulative iteration count with the update iteration count, and compares the inlier rate of the current first straight line model with a second preset threshold. If the comparison result shows that the cumulative iteration count is less than the update iteration count, and the inlier rate is not greater than the second preset threshold, then the process is repeated: two points are randomly selected from the selected point cluster to determine the second straight line model corresponding to the selected point cluster and subsequent steps, until the cumulative iteration count is greater than or equal to the update iteration count, or the inlier rate is greater than the second preset threshold, at which point the target straight line model corresponding to the selected point cluster is obtained. It is understood that the map generation system records the points selected in the point cluster and will not repeatedly select two identical points when selecting points in the point cluster the next time.

[0175] The map generation system in this embodiment determines the inlier rate of the first straight line model based on the inlier threshold and the distance from each point in the selected point cluster to the first straight line model; it calculates the number of update iterations based on the inlier rate and the confidence level of the first straight line model; it updates the first straight line model based on the number of update iterations, the inlier threshold, and the divergence threshold to obtain the target straight line model corresponding to the selected point cluster. By updating the straight line model through multiple iterations, the target straight line model corresponding to each point cluster is determined, making the accuracy of the road edges of the subsequently generated map higher.

[0176] Furthermore, such as Figure 4 As shown, based on the first and second embodiments of the map road edge line generation method of the present invention, a third embodiment of the map road edge line generation method of the present invention is proposed.

[0177] The third embodiment of the map road edge generation method differs from the first and second embodiments in that the step of determining the symmetric segmentation capability of the first straight line model and the second straight line model based on the interior point threshold and the divergence threshold, and updating the first straight line model based on the symmetric segmentation capability of the first straight line model and the second straight line model, includes:

[0178] Step S202361: Determine the first cumulative distance from the selected point cluster to the first straight line model for all points whose distance from the first straight line model is less than the preset multiple of the interior point threshold.

[0179] Step S202362: Determine the second cumulative distance from the selected point cluster to the first straight line model for all points whose distance from the first straight line model is greater than the divergence threshold.

[0180] Step S202363: Determine the symmetry segmentation capability of the first straight line model based on the first cumulative distance and the second cumulative distance;

[0181] In steps S202361 to S202363, the map generation system first determines a preset multiple threshold for points, and calculates the first distance from each point in the selected point cluster whose distance from the first straight line model is less than the preset multiple threshold to the first straight line model. All first distances are summed to obtain the first cumulative distance. The map generation system then determines a divergence threshold, and calculates the second distance from each point in the selected point cluster whose distance from the first straight line model is greater than the divergence threshold to the first straight line model. All second distances are summed to obtain the second cumulative distance. Finally, the symmetric segmentation capability of the first straight line model is determined based on the first and second cumulative distances. It should be noted that the smaller the first cumulative distance and the larger the second cumulative distance, the better the symmetric segmentation capability of the straight line model. Specifically, the preset multiple can be 1.5 times, 2 times, 3 times, etc., preferably 2 times. Figure 5 As shown, Figure 5 This diagram illustrates the relationship between the linear model and the interior point threshold and divergence threshold. Within the selected point cluster, there are points whose distance from the first linear model is greater than twice the interior point threshold but less than the divergence threshold; the region containing these points is the linear model. Figure 5 The blank areas in the model do not participate in determining the symmetry segmentation capability of the first straight line model.

[0182] Step S202364: Determine the third cumulative distance from the selected point cluster to the second straight line model for all points whose distance from the second straight line model is less than the preset multiple of the inner point threshold.

[0183] Step S202365, determining a fourth cumulative distance of all points in the selected point cluster having a distance greater than the divergence threshold to the second straight line model to the second straight line model;

[0184] Step S202366, determining the symmetry partitioning ability of the second straight line model based on the third cumulative distance and the fourth cumulative distance.

[0185] In steps S202364 to S202366, the process of determining the symmetry partitioning ability of the second straight line model is similar to that of the first straight line model, which will not be repeated here.

[0186] Further, before the step of updating the first straight line model according to the symmetry partitioning ability of the first straight line model and the second straight line model, the step comprises:

[0187] Step S202367, comparing the symmetry partitioning ability of the first straight line model and the symmetry partitioning ability of the second straight line model.

[0188] In this step, after determining the symmetry partitioning ability of the first straight line model and the symmetry partitioning ability of the second straight line model, the map generation system compares the symmetry partitioning ability of the first straight line model and the symmetry partitioning ability of the second straight line model, that is, compares the first cumulative distance corresponding to the first straight line model with the third cumulative distance corresponding to the second straight line model, and compares the second cumulative distance corresponding to the first straight line model with the fourth cumulative distance corresponding to the second straight line model.

[0189] Step S202368, if the symmetry partitioning ability of the first straight line model is better than the symmetry partitioning ability of the second straight line model, the first straight line model is not updated, and the step of determining the current cumulative iteration number and the current inlier rate of the first straight line model is performed.

[0190] In this step, after comparing the symmetry partitioning ability of the first straight line model and the symmetry partitioning ability of the second straight line model, if the comparison result is that the symmetry partitioning ability of the first straight line model is better than the symmetry partitioning ability of the second straight line model, that is, the first cumulative distance corresponding to the first straight line model is less than the third cumulative distance corresponding to the second straight line model, and the second cumulative distance corresponding to the first straight line model is greater than the fourth cumulative distance corresponding to the second straight line model, the first straight line model is not updated, that is, the first straight line model is not replaced by the second straight line model, and the step of determining the current cumulative iteration number and the current inlier rate of the first straight line model is performed.

[0191] It should be noted that the first straight line model is not updated, that is, the original first straight line model is maintained, at this time, the inlier rate and the update iteration number do not need to be recalculated, but the current cumulative iteration number is also increased by 1.

[0192] Further, the step of updating the first straight line model according to the symmetric segmentation ability of the first straight line model and the second straight line model comprises:

[0193] In step S202369, if the symmetric segmentation ability of the second straight line model is better than that of the first straight line model, the second straight line model is taken as the first straight line model to update the first straight line model.

[0194] In this step, after the map generation system compares the symmetric segmentation ability of the first straight line model and the symmetric segmentation ability of the second straight line model, if the comparison result is that the symmetric segmentation ability of the second straight line model is better than that of the first straight line model, that is, the third cumulative distance corresponding to the second straight line model is less than the first cumulative distance corresponding to the first straight line model, and the fourth cumulative distance corresponding to the second straight line model is greater than the second cumulative distance corresponding to the first straight line model, the second straight line model is taken as the first straight line model, the first straight line model is updated, and the current cumulative iteration number, the inlier rate of the current first straight line model and the subsequent steps are determined.

[0195] It should be noted that whether the map generation system updates or does not update the first straight line model, it is necessary to return to execute steps S202371 to S202374, that is, to determine the current cumulative iteration number and the inlier rate of the current first straight line model, to compare the cumulative iteration number with the update iteration number, and to compare the inlier rate of the current first straight line model with the second preset threshold, if the cumulative iteration number is greater than or equal to the update iteration number, or the inlier rate is greater than the second preset threshold, the current first straight line model is taken as the target straight line model corresponding to the selected point cluster, if the cumulative iteration number is less than the update iteration number, and the inlier rate is not greater than the second preset threshold, step S20231 and the subsequent steps are executed.

[0196] The map generation system of the present embodiment determines whether to update the first straight line model by calculating the symmetric segmentation ability of the first straight line model and the symmetric segmentation ability of the second straight line model and comparing the symmetric segmentation ability of the first straight line model with the symmetric segmentation ability of the second straight line model, which helps to improve the accuracy of straight line model updating, and further improves the accuracy of determining the target straight line model corresponding to each point cluster, so that the accuracy of the road edge line of the subsequently generated map is higher.

[0197] Further, as Figure 6As shown, based on the first embodiment to the third embodiment of the map road edge line generation method, the fourth embodiment of the map road edge line generation method is proposed.

[0198] The third embodiment of the map road edge line generation method is different from the first embodiment to the third embodiment of the map road edge line generation method in that the step of generating the map road edge line corresponding to the target road according to the target straight line model set comprises:

[0199] In step S301, the out points corresponding to each target straight line model in the target straight line model set are determined.

[0200] In step S302, the number of out points is compared with the third preset threshold.

[0201] In steps S301 to S302, after the map generation model determines the target straight line model set corresponding to the point cluster set, the out points in each point cluster corresponding to each target straight line model are determined according to each target straight line model in the target straight line model set and the divergence threshold, that is, the points in the point cluster with a distance greater than the divergence threshold from the target straight line model belong to the out points corresponding to the target straight line model. The map generation system counts the number of out points corresponding to all target straight line models, and compares the number of out points with the third preset threshold.

[0202] In step S303, if the number of out points is less than the third preset threshold, the map road edge line corresponding to the target road is generated according to the target straight line model set.

[0203] In step S304, if the number of out points is not less than the third preset threshold, new point cloud data is formed based on the out points, and the step of determining the point cluster set according to the point cloud data is re-executed based on the new point cloud data.

[0204] In steps S303 to S304, after the map generation system compares the number of out points with the third preset threshold, if the comparison result is that the number of out points is less than the third preset threshold, the map road edge line corresponding to the target road is generated according to all the target straight line model sets currently determined, that is, according to the geometric structure of the point cloud data itself, the target straight line models corresponding to each point cluster are connected to generate the map road edge line corresponding to the target road; if the comparison result is that the number of out points is not less than the third preset threshold, new point cloud data is formed based on all the out points, and the point cluster set is re-executed according to the point cloud data and the subsequent steps based on the new point cloud data until the target straight line model corresponding to each point cluster in the point cluster set corresponding to the new point cloud data is determined, and the number of remaining out points is less than the straight line model, at this time, the map road edge line corresponding to the target road is generated according to all the target straight line model sets.

[0205] The map generation system of the embodiment compares the number of all outliers with a third preset threshold to determine whether to determine a target straight line model corresponding to the point cloud data formed by the outliers, so as to utilize all the point cloud data as much as possible and improve the accuracy of the generated map road edge line.

[0206] Further, as shown in the fifth embodiment of the map road edge line generation method of the present application based on the first to fourth embodiments of the map road edge line generation method of the present application, the fifth embodiment of the map road edge line generation method of the present application is proposed. Figure 7 The fifth embodiment of the map road edge line generation method differs from the first to fourth embodiments of the map road edge line generation method in that the step of determining a target straight line model corresponding to each point cluster in the point cluster set to obtain a target straight line model set comprises:

[0207] In this step, after the map generation system determines the point cluster set corresponding to the point cloud data, a point cluster is selected from the point cluster set one by one, and a straight line model set corresponding to the selected point cluster is determined in the selected point cluster by determining a straight line model with two points.

[0208] In this step, after the map generation system determines the point cluster set corresponding to the point cloud data, a point cluster is selected from the point cluster set one by one, and a straight line model set corresponding to the selected point cluster is determined in the selected point cluster by determining a straight line model with two points.

[0209] In this step, the first preset threshold includes an inner point threshold and a divergence threshold. The map generation system determines the number of inner points corresponding to each straight line model in the straight line model set according to the inner point threshold, and determines the symmetric partitioning ability corresponding to each straight line model in the straight line model set according to the inner point threshold and the divergence threshold. It should be noted that the specific steps of determining the number of inner points corresponding to each straight line model and the symmetric partitioning ability are the same as those described above, and will not be described here.

[0210] In this step, the first preset threshold includes an inner point threshold and a divergence threshold. The map generation system determines the number of inner points corresponding to each straight line model in the straight line model set according to the inner point threshold, and determines the symmetric partitioning ability corresponding to each straight line model in the straight line model set according to the inner point threshold and the divergence threshold. It should be noted that the specific steps of determining the number of inner points corresponding to each straight line model and the symmetric partitioning ability are the same as those described above, and will not be described here.

[0211] In this step, the first preset threshold includes an inner point threshold and a divergence threshold. The map generation system determines the number of inner points corresponding to each straight line model in the straight line model set according to the inner point threshold, and determines the symmetric partitioning ability corresponding to each straight line model in the straight line model set according to the inner point threshold and the divergence threshold. It should be noted that the specific steps of determining the number of inner points corresponding to each straight line model and the symmetric partitioning ability are the same as those described above, and will not be described here.

[0212] In this step, the first preset threshold includes an inner point threshold and a divergence threshold. The map generation system determines the number of inner points corresponding to each straight line model in the straight line model set according to the inner point threshold, and determines the symmetric partitioning ability corresponding to each straight line model in the straight line model set according to the inner point threshold and the divergence threshold. It should be noted that the specific steps of determining the number of inner points corresponding to each straight line model and the symmetric partitioning ability are the same as those described above, and will not be described here.

[0213] In this step, after determining the number of inliers corresponding to each linear model in the linear model set and the symmetric segmentation capability corresponding to each linear model, the map generation system compares the number of inliers corresponding to all linear models, determines a linear model with the largest number of inliers in all linear models, and takes the linear model as the target linear model corresponding to the selected point cluster. If multiple linear models with the same and largest number of inliers are determined in all linear models, the symmetric segmentation capabilities corresponding to the multiple linear models are compared, a linear model with the optimal symmetric segmentation capability is determined in the multiple linear models, and the linear model is taken as the target linear model corresponding to the selected point cluster.

[0214] Step S207: Until the target linear model corresponding to each point cluster in the point cluster set is determined, the target linear model set corresponding to the point cluster set is obtained.

[0215] In this step, the map generation system sequentially performs the above steps S204 and S206 on each point cluster in the point cluster set, until the target linear model corresponding to each point cluster in the point cluster set is determined, and the target linear model set corresponding to the point cluster set is obtained.

[0216] The map generation system in this embodiment determines the point cluster set of the point cloud data, determines the target linear model of each point cluster in the point cluster set, and helps to construct the map road edge line corresponding to the target road through the target linear model, so that the map road edge line does not need to be manually made according to the point cloud data, and the generation efficiency of the map road edge line is improved.

[0217] As shown in Figure 8 The present application also provides a map road edge line generation device. The map road edge line generation device of the present application comprises:

[0218] The first determination module 101 is configured to obtain the point cloud data of the target road, and determine a point cluster set according to the point cloud data.

[0219] The second determination module 102 is configured to determine the target linear model corresponding to each point cluster in the point cluster set, and obtain a target linear model set.

[0220] The generation module 103 is configured to generate the map road edge line corresponding to the target road according to the target linear model set.

[0221] Further, the first determination module is further configured to:

[0222] The point cloud data is down-sampled, and the down-sampled point cloud data is clustered, the down-sampled point cloud data is segmented into multiple spatially discontinuous point clusters, and a point cluster set is obtained.

[0223] Further, the second determination module is further configured to:

[0224] determine a line model corresponding to each point cluster in the point cluster set, and update the line model corresponding to each point cluster to obtain a target line model set.

[0225] Further, the generating module is further configured to:

[0226] select a point cluster from the point cluster set in sequence, and randomly determine two points in the selected point cluster to determine a first line model corresponding to the selected point cluster;

[0227] update the first line model according to a first preset threshold and a distance between each point in the selected point cluster and the first line model to obtain a target line model corresponding to the selected point cluster;

[0228] until a target line model corresponding to each point cluster in the point cluster set is determined, a target line model set corresponding to the point cluster set is obtained.

[0229] Further, the generating module is further configured to:

[0230] determine an inlier rate corresponding to the first line model according to the inlier threshold and the distance between each point in the selected point cluster and the first line model;

[0231] calculate an update iteration number according to the confidence corresponding to the first line model and the inlier rate;

[0232] update the first line model based on the update iteration number, the inlier threshold and the divergence threshold to obtain a target line model corresponding to the selected point cluster.

[0233] Further, the generating module is further configured to:

[0234] randomly determine two points in the selected point cluster, and determine a second line model corresponding to the selected point cluster according to the selected two points;

[0235] calculate a distance between each point in the selected point cluster and the first line model and a distance between each point in the selected point cluster and the second line model respectively;

[0236] determine an inlier number of the first line model according to the distance between each point in the selected point cluster and the first line model and the inlier threshold, and determine an inlier number of the second line model according to the distance between each point in the selected point cluster and the second line model and the inlier threshold;

[0237] compare the inlier number of the second line model with the inlier number of the first line model;

[0238] if the number of inliers of the second straight line model is greater than the number of inliers of the first straight line model, the second straight line model is taken as the first straight line model to update the first straight line model;

[0239] if the number of inliers of the second straight line model is not greater than the number of inliers of the first straight line model, the symmetry segmentation ability of the first straight line model and the second straight line model is determined according to the inlier threshold and the divergence threshold, and the first straight line model is updated according to the symmetry segmentation ability of the first straight line model and the second straight line model.

[0240] Further, the generating module is further used for:

[0241] determining the current cumulative iteration number and the inlier rate of the current first straight line model;

[0242] comparing the cumulative iteration number with the update iteration number, and comparing the inlier rate of the current first straight line model with a second preset threshold;

[0243] if the cumulative iteration number is greater than or equal to the update iteration number, or the inlier rate is greater than the second preset threshold, the current first straight line model is taken as the target straight line model corresponding to the selected point cluster;

[0244] if the cumulative iteration number is less than the update iteration number, and the inlier rate is not greater than the second preset threshold, the step of randomly determining two points in the selected point cluster to determine the second straight line model corresponding to the selected point cluster is re-executed.

[0245] Further, the generating module is further used for:

[0246] determining the first cumulative distance of all points in the selected point cluster to the first straight line model, the distance of which to the first straight line model is less than a preset multiple of the inlier threshold;

[0247] determining the second cumulative distance of all points in the selected point cluster to the first straight line model, the distance of which to the first straight line model is greater than the divergence threshold;

[0248] determining the symmetry segmentation ability of the first straight line model based on the first cumulative distance and the second cumulative distance;

[0249] determining the third cumulative distance of all points in the selected point cluster to the second straight line model, the distance of which to the second straight line model is less than a preset multiple of the inlier threshold;

[0250] determining the fourth cumulative distance of all points in the selected point cluster to the second straight line model, the distance of which to the second straight line model is greater than the divergence threshold;

[0251] determine the symmetry segmentation capability of the second straight line model based on the third accumulated distance and the fourth accumulated distance.

[0252] Further, the generating module is further configured to:

[0253] compare the symmetry segmentation capability of the first straight line model with the symmetry segmentation capability of the second straight line model;

[0254] if the symmetry segmentation capability of the first straight line model is superior to the symmetry segmentation capability of the second straight line model, the first straight line model is not updated;

[0255] if the symmetry segmentation capability of the second straight line model is superior to the symmetry segmentation capability of the first straight line model, the second straight line model is taken as the first straight line model to update the first straight line model.

[0256] Further, the generating module is further configured to:

[0257] determine the outlier corresponding to each target straight line model in the target straight line model set;

[0258] compare the number of outliers with a third preset threshold value;

[0259] if the number of outliers is less than the third preset threshold value, the map road edge line corresponding to the target road is generated according to the target straight line model set;

[0260] if the number of outliers is not less than the third preset threshold value, new point cloud data is constituted based on the outliers, and the step of determining the point cluster set according to the point cloud data is re-executed based on the new point cloud data.

[0261] The application further provides a map road edge line generation system.

[0262] The map road edge line generation system comprises a memory, a processor and a map road edge line generation program stored in the memory and executable on the processor, and the map road edge line generation program, when executed by the processor, implements the steps of the map road edge line generation method.

[0263] The method implemented when the map road edge line generation program executable on the processor is executed can refer to the embodiments of the map road edge line generation method of the application, and will not be repeated here.

[0264] The application further provides a readable storage medium.

[0265] The map road edge generation program stored on the readable storage medium of the present application is executed by a processor to implement the steps of the map road edge generation method as described above.

[0266] The method implemented by the map road edge generation program running on the processor can refer to the embodiments of the map road edge generation method of the present application, which will not be described here.

[0267] It should be noted that in this document, the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such a process, method, article or system. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0268] The above-mentioned embodiment numbers of the present application are only for description, not representing the advantages or disadvantages of the embodiments.

[0269] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and necessary general hardware platforms, of course, they can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, an optical disk) as described above, and includes a number of instructions to make a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) execute the methods described in various embodiments of the present application.

[0270] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, is also included in the patent protection scope of the present application.

Claims

1. A method for generating map road edges, characterized in that, The method for generating map road edges includes the following steps: Acquire point cloud data of the target road, and determine the point cluster set based on the point cloud data; Determine the target straight line model corresponding to each point cluster in the point cluster set to obtain the target straight line model set; Generate the map road edge lines corresponding to the target road based on the target straight line model set; The step of determining the target straight line model corresponding to each point cluster in the point cluster set to obtain the target straight line model set includes: Select one point cluster from the set of point clusters in sequence, and determine the set of straight line models corresponding to the selected point cluster by using two points to determine a straight line model in the selected point cluster; Based on a first preset threshold, determine the number of interior points and symmetry segmentation capability of each straight line model in the set of straight line models; Based on the number of interior points and the symmetric segmentation capability, the target straight line model corresponding to the selected point cluster is determined in the set of straight line models; The process continues until the target straight line model corresponding to each point cluster in the point cluster set is determined, thus obtaining the target straight line model set corresponding to the point cluster set.

2. The map road edge generation method as described in claim 1, characterized in that, The step of determining the point cluster set based on the point cloud data includes: The point cloud data is downsampled, and the downsampled point cloud data is clustered to divide the downsampled point cloud data into multiple spatially discontinuous point clusters, thus obtaining a point cluster set.

3. The map road edge generation method as described in claim 1, characterized in that, The step of determining the target straight line model corresponding to each point cluster in the point cluster set to obtain the target straight line model set further includes: Select a point cluster from the set of point clusters in sequence, and randomly determine two points in the selected point cluster to determine the first straight line model corresponding to the selected point cluster; The first straight line model is updated based on the first preset threshold and the distance from each point in the selected point cluster to the first straight line model, so as to obtain the target straight line model corresponding to the selected point cluster. The process continues until the target straight line model corresponding to each point cluster in the point cluster set is determined, thus obtaining the target straight line model set corresponding to the point cluster set.

4. The map road edge generation method as described in claim 3, characterized in that, The first preset threshold includes an interior point threshold and a divergence threshold. The step of updating the first straight line model based on the first preset threshold and the distance from each point in the selected point cluster to the first straight line model to obtain the target straight line model corresponding to the selected point cluster includes: Based on the inlier threshold and the distance from each point in the selected point cluster to the first straight line model, the inlier rate corresponding to the first straight line model is determined. The number of update iterations is calculated based on the confidence level and the inlier rate corresponding to the first linear model. The first straight line model is updated based on the number of update iterations, the interior point threshold, and the divergence threshold to obtain the target straight line model corresponding to the selected point cluster.

5. The map road edge generation method as described in claim 4, characterized in that, The step of updating the first straight line model based on the number of update iterations, the interior point threshold, and the divergence threshold to obtain the target straight line model corresponding to the selected point cluster includes: Two points are randomly selected from the selected point cluster, and the second straight line model corresponding to the selected point cluster is determined based on the two selected points. Calculate the distance from each point in the selected point cluster to the first straight line model and the distance from each point in the selected point cluster to the second straight line model, respectively. The number of interior points of the first straight line model is determined based on the distance from each point in the selected point cluster to the first straight line model and the interior point threshold; and the number of interior points of the second straight line model is determined based on the distance from each point in the selected point cluster to the second straight line model and the interior point threshold. Compare the number of interior points of the second straight line model with the number of interior points of the first straight line model; If the number of interior points of the second straight line model is greater than the number of interior points of the first straight line model, then the second straight line model is used as the first straight line model, and the first straight line model is updated. If the number of interior points of the second straight line model is not greater than the number of interior points of the first straight line model, then the symmetric segmentation capability of the first straight line model and the second straight line model is determined according to the interior point threshold and the divergence threshold, and the first straight line model is updated according to the symmetric segmentation capability of the first straight line model and the second straight line model. Based on the update result of the first straight line model, the target straight line model corresponding to the selected point cluster is determined.

6. The map road edge generation method as described in claim 5, characterized in that, The step of determining the target straight line model corresponding to the selected point cluster based on the update result of the first straight line model includes: Determine the current cumulative iteration count and the interior point rate of the current first straight line model; The cumulative number of iterations is compared with the number of update iterations, and the inlier rate of the current first straight line model is compared with the second preset threshold. If the cumulative number of iterations is greater than or equal to the number of update iterations, or the inlier rate is greater than the second preset threshold, then the current first straight line model is used as the target straight line model corresponding to the selected point cluster. If the cumulative iteration count is less than the update iteration count, and the interior point rate is not greater than the second preset threshold, then the following step is executed again: randomly determine two points in the selected point cluster to determine the second straight line model corresponding to the selected point cluster.

7. The map road edge generation method as described in claim 5, characterized in that, The step of determining the symmetric segmentation capability of the first linear model and the second linear model based on the interior point threshold and the divergence threshold includes: Determine the first cumulative distance from the selected point cluster to the first straight line model for all points whose distance from the first straight line model is less than a preset multiple of the in-point threshold. Determine the second cumulative distance from the selected point cluster to the first straight line model for all points whose distance from the first straight line model is greater than the divergence threshold; The symmetric segmentation capability of the first straight line model is determined based on the first cumulative distance and the second cumulative distance; The third cumulative distance from the selected point cluster to the second straight line model is determined for all points whose distance from the second straight line model is less than the preset multiple of the interior point threshold. The fourth cumulative distance from the selected point cluster to the second straight line model is determined for all points whose distance from the second straight line model is greater than the divergence threshold. The symmetric segmentation capability of the second linear model is determined based on the third cumulative distance and the fourth cumulative distance.

8. The map road edge generation method as described in claim 5, characterized in that, Before the step of updating the first straight line model based on the symmetric segmentation capability of the first straight line model and the second straight line model, the method further includes: The symmetric segmentation capabilities of the first linear model and the second linear model are compared. If the symmetric segmentation capability of the first linear model is better than that of the second linear model, then the first linear model is not updated, and the following steps are performed: determine the current cumulative iteration count and the current inlier rate of the first linear model; The step of updating the first straight line model based on the symmetric segmentation capability of the first straight line model and the second straight line model includes: If the symmetric segmentation capability of the second straight line model is better than that of the first straight line model, then the second straight line model is used as the first straight line model, and the first straight line model is updated.

9. The map road edge generation method as described in claim 1, characterized in that, The step of generating the map road edge line corresponding to the target road based on the target straight line model set includes: Determine the exterior point corresponding to each target straight line model in the set of target straight line models; The number of the outer points is compared with a third preset threshold. If the number of external points is less than the third preset threshold, then the map road edge line corresponding to the target road is generated according to the target straight line model set; If the number of external points is not less than the third preset threshold, then new point cloud data is constructed based on the external points, and the step of determining the point cluster set based on the new point cloud data is executed again.

10. A map road edge generation device, characterized in that, The map road edge generation device includes: The first determining module is used to acquire point cloud data of the target road and determine a set of point clusters based on the point cloud data; The second determining module is used to determine the target straight line model corresponding to each point cluster in the point cluster set, thereby obtaining a target straight line model set. The generation module is used to generate the map road edge lines corresponding to the target road based on the target straight line model set; The first determining module is further configured to sequentially select a point cluster from the point cluster set, and determine the set of straight line models corresponding to the selected point cluster by determining a straight line model by two points in the selected point cluster; determine the number of interior points and symmetry segmentation capability corresponding to each straight line model in the set of straight line models according to a first preset threshold; determine the target straight line model corresponding to the selected point cluster in the set of straight line models based on the number of interior points and the symmetry segmentation capability; until the target straight line model corresponding to each point cluster in the set of point clusters is determined, the set of target straight line models corresponding to the point cluster set is obtained.

11. A map road edge generation system, characterized in that, The map road edge generation system includes: a memory, a processor, and a map road edge generation program stored in the memory and executable on the processor. When the map road edge generation program is executed by the processor, it implements the steps of the map road edge generation method as described in any one of claims 1 to 9.

12. A readable storage medium, characterized in that, The readable storage medium stores a map road edge generation program, which, when executed by a processor, implements the steps of the map road edge generation method as described in any one of claims 1 to 9.

Images