A vehicle control method and device, a vehicle, a storage medium, and a program product.

Abstract

This application provides a vehicle control method and device, a vehicle, a storage medium, and a program product; the method includes: in response to a vehicle entering a parking mode, determining the ROI area of ​​the vehicle; determining point cloud clusters corresponding to sensing data located in the ROI area; the sensing data includes: visual point clouds, radar point clouds, and ultrasonic point clouds; for each point cloud cluster, determining the circumscribed rectangle of the point cloud cluster; determining a target point cloud cluster from the point cloud clusters based on the point cloud density of the point cloud clusters and / or the geometric features of the circumscribed rectangle corresponding to the point cloud clusters; determining the attributes of the target point cloud clusters and determining the location of obstacles; generating a parking path based on the obstacle location to control the vehicle to park based on the parking path.

Description

Technical Field

[0001] This application relates to active safety technology, specifically to a vehicle control method and device, a vehicle, a storage medium, and a program product. Background Technology

[0002] With the continuous evolution of autonomous driving technology, automatic parking functions place extremely high demands on the accuracy and robustness of the vehicle's near-field environmental perception. In low-speed parking scenarios, vehicles typically rely on multimodal sensors for environmental modeling. However, existing sensors all have physical limitations; for example, although ultrasonic radar has accurate ranging, it suffers from severe beam divergence, extremely low angular resolution, and is susceptible to sidelobe interference, resulting in false obstacle warnings. Summary of the Invention

[0003] This application provides a vehicle control method and device, a vehicle, a storage medium, and a computer program product, which are beneficial for suppressing false obstacle alarms caused by ground water stains, illusory lines, and sidelobe interference.

[0004] In a first aspect, embodiments of this application provide a vehicle control method, the method comprising: in response to a vehicle entering a parking mode, determining the ROI area of ​​the vehicle; determining point cloud clusters corresponding to sensing data located in the ROI area; the sensing data including: visual point clouds, radar point clouds, and ultrasonic point clouds; for each point cloud cluster, determining the circumscribed rectangle of the point cloud cluster; determining a target point cloud cluster from the point cloud clusters based on the point cloud density of the point cloud clusters and / or the geometric features of the circumscribed rectangle corresponding to the point cloud clusters; determining the attributes of the target point cloud clusters and determining the location of obstacles; generating a parking path based on the location of the obstacles, so as to control the vehicle to park based on the parking path.

[0005] It is understood that in the vehicle control method provided in this application embodiment, when the vehicle enters parking mode, the Region of Interest (ROI) of the vehicle is determined; and the point cloud clusters corresponding to the perceived data located within the ROI are determined; for each point cloud cluster, the circumscribed rectangle of the point cloud cluster is determined; thus, based on the point cloud density and / or the geometric features of the circumscribed rectangle, the point cloud clusters are filtered to determine the target point cloud clusters, and the attributes of the target point cloud clusters are determined; based on this, the obstacle positions are determined, a parking path is generated, and the vehicle is controlled to park based on the parking path. In this way, filtering the point cloud clusters by the outline of the circumscribed rectangle and the density of the point cloud clusters helps to suppress false obstacle alarms caused by ground water stains, phantom lines, and sidelobe interference, thereby making the determined obstacle positions more reliable and making parking more intelligent.

[0006] In some embodiments, determining a target point cloud cluster from the point cloud cluster based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster includes at least one of the following: removing the point cloud cluster if the point cloud density of the point cloud cluster is less than or equal to a first density threshold; removing the point cloud cluster if the aspect ratio of the bounding rectangle corresponding to the point cloud cluster is greater than or equal to an aspect ratio threshold; and removing the point cloud cluster if the area of ​​the bounding rectangle corresponding to the point cloud cluster is less than or equal to an area threshold.

[0007] It is understood that in the vehicle control method provided in this application embodiment, point cloud clusters with a point cloud density less than or equal to a first density threshold are discarded; point cloud clusters with an aspect ratio of their circumscribed rectangle greater than or equal to an aspect ratio threshold are discarded; and point cloud clusters with an area of ​​their circumscribed rectangle less than or equal to an area threshold are discarded. Thus, filtering point cloud clusters based on the geometric characteristics of their circumscribed rectangles and the density of the point cloud clusters helps to eliminate invalid scattering points and effectively suppress false obstacle alarms caused by ground water stains, phantom lines, and sidelobe interference.

[0008] In some embodiments, determining the point cloud clusters corresponding to the sensing data located in the ROI region includes: performing dilation and / or erosion operations on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud; superimposing the processed radar point cloud and the processed visual point cloud to obtain a superimposed point cloud; and clustering the processed ultrasonic point cloud and the superimposed point cloud to obtain multiple point cloud clusters.

[0009] It is understood that in the vehicle control method provided in the embodiments of this application, firstly, the visual point cloud, radar point cloud and ultrasonic point cloud located in the ROI area are subjected to dilation and / or erosion operations, and then the processed radar point cloud and visual point cloud are superimposed to obtain a superimposed point cloud. The processed ultrasonic point cloud and superimposed point cloud are clustered respectively to obtain multiple point cloud clusters. This is beneficial to eliminate isolated noise and ground reflection interference.

[0010] In some embodiments, determining the attributes of the target point cloud cluster further includes: predicting the vehicle's driving trajectory based on the acquired vehicle state information; projecting the target point cloud cluster onto the driving trajectory if the point cloud density of the target point cloud cluster is less than or equal to a second density threshold; wherein the second density threshold is greater than a first density threshold; and determining the tangent direction of the projection point of the target point cloud cluster on the driving trajectory as the orientation of the target point cloud cluster.

[0011] It is understood that in the vehicle control method provided in this application embodiment, the vehicle's driving trajectory is predicted based on the acquired vehicle state information; when the point cloud density of the target point cloud cluster is less than or equal to a second density threshold, the target point cloud cluster is projected onto the driving trajectory; and the tangent direction of the projection point of the target point cloud cluster on the driving trajectory is determined as the orientation of the target point cloud cluster. Thus, when the point cloud density of the target point cloud cluster is lower than a certain threshold, it is considered a missed detection, and it is projected onto the predicted driving trajectory, with the orientation of the target point cloud cluster determined as the tangent direction of the projection point onto the predicted driving trajectory, thereby improving obstacle avoidance rate and driving safety.

[0012] In some embodiments, determining the circumscribed rectangle of the point cloud cluster includes: determining the convex hull polygon of the point cloud cluster; using a rotating caliper algorithm, traversing each edge of the convex hull polygon, constructing a corresponding circumscribed rectangle for each edge, and determining the area of ​​each circumscribed rectangle; and determining the circumscribed rectangle with the smallest area as the circumscribed rectangle of the point cloud cluster.

[0013] It is understood that in the vehicle control method provided in this application embodiment, the convex hull polygon of the point cloud cluster is determined; using the rotating caliper algorithm, each edge of the convex hull polygon is traversed, and a corresponding circumscribed rectangle is constructed for each edge; the circumscribed rectangle with the smallest area among multiple circumscribed rectangles is determined as the circumscribed rectangle of the point cloud cluster. This is beneficial for using the rotating caliper method to correct the geometric contour of the point cloud cluster, thereby effectively eliminating the positioning drift problem caused by beam divergence.

[0014] In some embodiments, the method further includes: smoothing the updated target point cloud cluster using a Kalman filter algorithm.

[0015] It is understood that in the vehicle control method provided in the embodiments of this application, the Kalman filter algorithm is used to smooth the updated target point cloud cluster, which is beneficial to eliminating the positioning drift caused by beam divergence.

[0016] Secondly, embodiments of this application provide a vehicle control device, comprising: a first determining module configured to determine the ROI area of ​​the vehicle in response to the vehicle entering a parking mode; a second determining module configured to determine point cloud clusters corresponding to sensing data located in the ROI area; the sensing data including: visual point clouds, radar point clouds, and ultrasonic point clouds; a third determining module configured to determine the circumscribed rectangle of each point cloud cluster; a fourth determining module configured to determine a target point cloud cluster from the point cloud clusters based on the point cloud density of the point cloud clusters and / or the geometric features of the circumscribed rectangle corresponding to the point cloud clusters; a fifth determining module configured to determine the attributes of the target point cloud clusters and determine the location of obstacles; and a control module configured to generate a parking path based on the location of the obstacles to control the vehicle to park based on the parking path.

[0017] Thirdly, embodiments of this application provide a vehicle, including: a memory and a processor; the memory stores a computer program that can run on the processor, and the processor executes the program to implement the vehicle control method described in the first aspect.

[0018] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the vehicle control method described in the first aspect.

[0019] Fifthly, embodiments of this application provide a computer program product, including a computer program or instructions, which, when executed by a processor, implement the vehicle control method described in the first aspect. Attached Figure Description

[0020] Figure 1 A schematic diagram of the implementation process of a vehicle control method provided in this application embodiment. Figure 1 ; Figure 2 A schematic diagram illustrating the implementation process of determining a point cloud cluster, provided in an embodiment of this application; Figure 3 A schematic diagram illustrating the implementation process of determining the circumscribed rectangle of a point cloud cluster, provided in an embodiment of this application; Figure 4 A schematic diagram illustrating the implementation process for determining the attributes of a target point cloud cluster, provided in an embodiment of this application; Figure 5 A schematic diagram of the implementation process of a vehicle control method provided in this application embodiment. Figure 2 ; Figure 6 A schematic diagram of the implementation process of a ROI region provided in this application embodiment. Figure 1 ; Figure 7A schematic diagram of the implementation process of a ROI region provided in this application embodiment. Figure 2 ; Figure 8 A schematic diagram of a vehicle control device provided in an embodiment of this application; Figure 9 This is a structural schematic diagram of a vehicle provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described below in conjunction with the accompanying drawings. The embodiments described below are only some embodiments of this application, not all embodiments. Therefore, the described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] In the following description, references to “some embodiments” or “other embodiments” describe a subset of all possible embodiments. However, it is understood that “some embodiments” or “other embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0024] In the following description, the terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first" and "second" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0026] With the continuous evolution of autonomous driving technology, the Automated Parking Assist (APA) function places extremely high demands on the accuracy and robustness of the vehicle's near-field environmental perception. In low-speed parking scenarios, vehicles typically rely on multimodal sensors for environmental modeling. However, existing sensors all have physical limitations; for example, while ultrasonic radar (USS) has accurate ranging, it suffers from severe beam divergence and extremely low angular resolution, making it susceptible to sidelobe interference that generates false obstacle alarms, and it cannot accurately determine the geometry of obstacles; millimeter-wave radar has sparse point clouds when detecting at close range, making it difficult to construct accurate contours; and visual sensors are easily affected by lighting conditions and lack of texture.

[0027] In related technologies, simple grid map overlay or nearest-neighbor data association algorithms are often used to achieve the fusion of multi-sensor data. However, these methods fail to fully utilize the vehicle's own kinematic prior information. This leads to a serious waste of computational resources in complex parking scenarios (such as narrow road passage, angled parking, and irregular curbs), and it is also very easy to misjudge the fan-shaped scattering of ultrasonic waves as physical obstacles or the curb as a wall, resulting in target attribute recognition errors and positioning deviations, which seriously affect parking safety and experience.

[0028] In view of this, this application provides a vehicle control method. Figure 1 A schematic diagram of the implementation process of a vehicle control method provided in this application embodiment. Figure 1 ,like Figure 1 As shown, the method includes steps 101 to 106: Step 101: In response to the vehicle entering parking mode, determine the ROI area of ​​the vehicle; Step 102: Determine the point cloud clusters corresponding to the sensing data located in the ROI region; the sensing data includes: visual point cloud, radar point cloud and ultrasonic point cloud; Step 103: For each point cloud cluster, determine the bounding rectangle of the point cloud cluster; Step 104: Based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster, determine the target point cloud cluster from the point cloud cluster. Step 105: Determine the attributes of the target point cloud cluster and determine the location of the obstacle; Step 106: Generate a parking path based on the location of the obstacle, so as to control the vehicle to park based on the parking path.

[0029] It is understood that in the vehicle control method provided in this application embodiment, when the vehicle enters parking mode, the Region of Interest (ROI) of the vehicle is determined; and the point cloud clusters corresponding to the perceived data located within the ROI are determined; for each point cloud cluster, the circumscribed rectangle of the point cloud cluster is determined; thus, based on the point cloud density and / or the geometric features of the circumscribed rectangle, the point cloud clusters are filtered to determine the target point cloud clusters, and the attributes of the target point cloud clusters are determined; based on this, the obstacle positions are determined, a parking path is generated, and the vehicle is controlled to park based on the parking path. In this way, filtering the point cloud clusters by the outline of the circumscribed rectangle and the density of the point cloud clusters helps to suppress false obstacle alarms caused by ground water stains, phantom lines, and sidelobe interference, thereby making the determined obstacle positions more reliable and making parking more intelligent.

[0030] The following sections will describe further optional implementation methods for each of the above steps, as well as related terms.

[0031] In step 101, in response to the vehicle entering parking mode, the ROI area of ​​the vehicle is determined.

[0032] In some embodiments, determining the ROI region of the vehicle in response to the vehicle entering a parking mode includes: generating the ROI region of the vehicle based on acquired vehicle state information in response to the vehicle entering a parking mode. Further, in some embodiments, generating the ROI region of the vehicle based on the acquired vehicle state information includes: determining the turning radius of the vehicle using an Ackermann steering geometry model based on the acquired vehicle state information; and generating the ROI region of the vehicle based on the turning radius.

[0033] It should be understood that the vehicle's state information is not limited in this embodiment. In some embodiments, the vehicle's state information includes, but is not limited to, one or more of the following: the vehicle's current position (x, y), heading. And the current speed v and front wheel steering angle The vehicle's wheelbase L, vehicle length, vehicle width, and vehicle gears.

[0034] In some embodiments, generating the ROI region of the vehicle based on the turning radius further includes: determining the ROI region of the vehicle based on the turning radius and the vehicle's gear position. Further, in some embodiments, determining the ROI region of the vehicle based on the turning radius and the vehicle's gear position includes: when the vehicle is in drive, determining the ROI region of the vehicle located in front of the vehicle based on the turning radius; and when the vehicle is in reverse, determining the ROI region of the vehicle located behind the vehicle based on the turning radius.

[0035] For example, in one possible implementation, when the vehicle is in forward gear, a forward fan-shaped or rectangular trajectory ROI region is generated along the vehicle's orientation, starting from the center of the front suspension; when the vehicle is in reverse gear, a backward trajectory ROI region is generated along the opposite orientation, starting from the center of the rear axle; the width of the ROI region is dynamically expanded according to the vehicle's physical width and a preset safety distance.

[0036] It is understood that in this embodiment of the application, the ROI area of ​​the vehicle is determined and the data within the ROI area is processed; in this way, dense calculation across the entire field of view is avoided, the load on the central processing unit (CPU) of the embedded controller is greatly reduced, and the real-time requirements of automatic parking are met.

[0037] In step 102, the point cloud clusters corresponding to the sensing data located in the ROI region are determined; the sensing data includes: visual point clouds, radar point clouds and ultrasonic point clouds.

[0038] In some embodiments, Figure 2 This application provides a schematic diagram of an implementation process for determining point cloud clusters, as illustrated in the embodiments of this application. Figure 2 As shown, determining the point cloud clusters corresponding to the sensing data located in the ROI region includes the following steps 201 to 203: Step 201: Perform dilation and / or erosion operations on the visual point cloud, radar point cloud and ultrasonic point cloud located in the ROI region to obtain the processed visual point cloud, processed radar point cloud and processed ultrasonic point cloud. Step 202: Overlay the processed radar point cloud and the processed visual point cloud to obtain an overlay point cloud; Step 203: Cluster the processed ultrasonic point cloud and the superimposed point cloud respectively to obtain multiple point cloud clusters.

[0039] It is understood that in the vehicle control method provided in the embodiments of this application, firstly, the visual point cloud, radar point cloud and ultrasonic point cloud located in the ROI area are subjected to dilation and / or erosion operations, and then the processed radar point cloud and visual point cloud are superimposed to obtain a superimposed point cloud. The processed ultrasonic point cloud and superimposed point cloud are clustered respectively to obtain multiple point cloud clusters. This is beneficial to eliminate isolated noise and ground reflection interference.

[0040] It should be understood that in the embodiments of this application, the visual point cloud, radar point cloud, and ultrasonic point cloud should be time-aligned. In some embodiments, the clocks of the visual point cloud acquisition device (visual sensor), the radar point cloud acquisition device (radar sensor), and the ultrasonic point cloud acquisition device (ultrasonic sensor) are synchronized to the same Global Positioning System (GPS) clock via the Precision Time Protocol (PTP). In other embodiments, the visual point cloud acquisition device, the radar point cloud acquisition device, and the ultrasonic point cloud acquisition device assign a "time tag" to a frame of data at the moment of acquisition, and subsequent processing uses this tag to find and associate data belonging to the same time. Furthermore, in some embodiments, the visual point cloud, radar point cloud, and ultrasonic point cloud are time-aligned using a time alignment algorithm.

[0041] It should be understood that, in the embodiments of this application, the visual point cloud, the radar point cloud, and the ultrasonic point cloud should be spatially aligned. Spatial alignment means that the visual point cloud acquisition device, the radar point cloud acquisition device, and the ultrasonic point cloud acquisition device acquire data from the same physical location.

[0042] In some embodiments, performing dilation and / or erosion operations on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud includes one of the following: performing dilation operations on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud; performing erosion operations on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud; performing dilation and erosion operations sequentially on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud; performing erosion and dilation operations sequentially on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud.

[0043] In some embodiments, superimposing the processed radar point cloud and the processed visual point cloud to obtain a superimposed point cloud includes: matching the processed radar point cloud and the processed visual point cloud, and superimposing the matched processed radar point cloud and the processed visual point cloud to obtain a superimposed point cloud. Further, in some embodiments, matching the processed radar point cloud and the processed visual point cloud includes: determining that the processed radar point cloud and the processed visual point cloud match if the first distance between the processed radar point cloud and the processed visual point cloud is less than a first distance threshold.

[0044] It should be understood that the first distance threshold is not limited in the embodiments of this application. In some embodiments, the first distance threshold is preset based on experience. For example, in one possible implementation, the first distance threshold is 1.8 meters.

[0045] In other embodiments, the step of superimposing the processed radar point cloud and the processed visual point cloud to obtain a superimposed point cloud includes: mapping the processed radar point cloud and the processed visual point cloud to the same three-dimensional voxel space to generate a corresponding binary voxel map; performing a bitwise AND operation on the binary voxel map, retaining the voxels occupied by both data sources simultaneously, and outputting a fused voxel map; and obtaining the superimposed point cloud based on the fused voxel map.

[0046] In other embodiments, determining the point cloud clusters corresponding to the sensing data located in the ROI region includes: performing dilation operations on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region to obtain dilated visual point clouds, dilated radar point clouds, and dilated ultrasonic point clouds, respectively; performing erosion operations on the dilated visual point cloud, the dilated radar point cloud, and the dilated ultrasonic point cloud, respectively to obtain eroded visual point clouds, eroded radar point clouds, and eroded ultrasonic point clouds, respectively; performing a bitwise AND operation on the dilated visual point cloud and the eroded visual point cloud to obtain a superimposed visual point cloud; performing a bitwise AND operation on the dilated radar point cloud and the eroded radar point cloud to obtain a superimposed radar point cloud; performing a bitwise AND operation on the dilated ultrasonic point cloud and the eroded ultrasonic point cloud to obtain a superimposed ultrasonic point cloud; superimposing the superimposed visual point cloud and the superimposed radar point cloud to obtain a superimposed point cloud; and clustering the superimposed ultrasonic point cloud and the superimposed point cloud to obtain multiple point cloud clusters.

[0047] It should be understood that, in this embodiment, ultrasonic point clouds are retained because they are more reliable. Radar sensors are susceptible to distance and light reflection; visual sensors are susceptible to lighting conditions and texture loss. Therefore, the reliability of radar and visual point clouds is relatively low. Thus, the processed radar and visual point clouds are superimposed to obtain a more reliable superimposed point cloud.

[0048] In step 103, for each point cloud cluster, the circumscribed rectangle of the point cloud cluster is determined.

[0049] In some embodiments, Figure 3 This application provides a schematic diagram of the implementation process for determining the circumscribed rectangle of a point cloud cluster, as shown in the embodiments of this application. Figure 3 As shown, determining the circumscribed rectangle of the point cloud cluster includes the following steps 301 to 303: Step 301: Determine the convex hull polygon of the point cloud cluster; Step 302: Using the rotating caliper algorithm, traverse each edge of the convex hull polygon, construct a corresponding circumscribed rectangle for each edge, and determine the area of ​​each circumscribed rectangle; Step 303: Determine the bounding rectangle with the smallest area as the bounding rectangle of the point cloud cluster.

[0050] It is understood that in the vehicle control method provided in this application embodiment, the convex hull polygon of the point cloud cluster is determined; using the rotating caliper algorithm, each edge of the convex hull polygon is traversed, and a corresponding circumscribed rectangle is constructed for each edge; the circumscribed rectangle with the smallest area among multiple circumscribed rectangles is determined as the circumscribed rectangle of the point cloud cluster. This is beneficial for using the rotating caliper method to correct the geometric contour of the point cloud cluster, thereby effectively eliminating the positioning drift problem caused by beam divergence.

[0051] It should be understood that, in the embodiments of this application, the convex hull polygon of the point cloud cluster is not limited. In some embodiments, the convex hull polygon refers to the polygon obtained by ignoring the internal holes and depressions of the point cloud cluster and retaining only the outline points of the outermost ring of the point cloud cluster.

[0052] In some embodiments, the step of using the rotating caliper algorithm to traverse each edge of the convex hull polygon, construct a corresponding circumscribed rectangle for each edge, and determine the area of ​​each circumscribed rectangle includes: taking the first edge of the convex hull polygon as one edge of the circumscribed rectangle, drawing the smallest circumscribed rectangle that can enclose the convex hull polygon; calculating the area of ​​the circumscribed rectangle; discarding the first edge of the convex hull polygon and moving to the next edge of the convex hull polygon (rotating the caliper counterclockwise); iteratively executing the above steps until each edge of the convex hull polygon has undergone the above operation.

[0053] It should be understood that the point cloud cluster is not limited in the embodiments of this application. In some embodiments, the point cloud cluster is obtained by grouping discrete points within the ROI region according to their distance, with points that are close together being classified into the same cluster, representing the same obstacle (e.g., a vehicle or a pedestrian).

[0054] In step 104, a target point cloud cluster is determined from the point cloud cluster based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster.

[0055] It should be understood that the geometric features of the circumscribed rectangle are not limited in this embodiment. In some embodiments, the geometric features of the circumscribed rectangle include, but are not limited to, at least one of the following: the aspect ratio of the circumscribed rectangle, the area of ​​the circumscribed rectangle, and the aspect ratio of the circumscribed rectangle.

[0056] In some embodiments, determining a target point cloud cluster from the point cloud cluster based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster includes at least one of the following: removing the point cloud cluster if the point cloud density of the point cloud cluster is less than or equal to a first density threshold; removing the point cloud cluster if the aspect ratio of the bounding rectangle corresponding to the point cloud cluster is greater than or equal to an aspect ratio threshold; and removing the point cloud cluster if the area of ​​the bounding rectangle corresponding to the point cloud cluster is less than or equal to an area threshold.

[0057] It is understood that in the vehicle control method provided in this application embodiment, point cloud clusters with a point cloud density less than or equal to a first density threshold are discarded; point cloud clusters with an aspect ratio of their circumscribed rectangle greater than or equal to an aspect ratio threshold are discarded; and point cloud clusters with an area of ​​their circumscribed rectangle less than or equal to an area threshold are discarded. Thus, filtering point cloud clusters based on the geometric characteristics of their circumscribed rectangles and the density of the point cloud clusters helps to eliminate invalid scattering points and effectively suppress false obstacle alarms caused by ground water stains, phantom lines, and sidelobe interference.

[0058] Furthermore, in some embodiments, the retained point cloud clusters are determined as the target point cloud clusters. It should be understood that, in the embodiments of this application, point cloud clusters that are not eliminated are the target point cloud clusters.

[0059] It should be understood that the point cloud density of the point cloud cluster is not limited in the embodiments of this application. In some embodiments, the point cloud density of the point cloud cluster is determined based on the number of point clouds in the cluster and the enclosing area of ​​the cluster. In other embodiments, the point cloud density of the point cloud cluster is determined by: for each point cloud in the cluster, determining the second distance between each point cloud in the cluster and the point cloud within a second distance threshold; determining a first average distance based on a plurality of second distances; and determining the point cloud density of the cluster based on the first average distance. In still other embodiments, the point cloud density of the point cloud cluster is determined by: for each point cloud in the cluster, determining the third distance between each of the K nearest point clouds in the cluster and the point cloud; determining a second average distance based on a plurality of third distances; and determining the point cloud density of the cluster based on the second average distance.

[0060] It should be understood that in the embodiments of this application, the smaller the average distance, the greater the point cloud density of the point cloud cluster; the greater the average distance, the smaller the point cloud density of the point cloud cluster.

[0061] In this embodiment, the second distance threshold and K are not limited. In some embodiments, the second distance threshold and K are preset based on experience. For example, in one possible implementation, the second distance is 0.2 meters; and K is 3.

[0062] It should be understood that, in this embodiment, the first density threshold, the aspect ratio threshold, and the area threshold are not limited. In some embodiments, the first density threshold, the aspect ratio threshold, and the area threshold are preset based on experience. In this embodiment, if the point cloud density of a point cloud cluster is less than or equal to the first density threshold, the aspect ratio of the bounding rectangle corresponding to the point cloud cluster is greater than or equal to the aspect ratio threshold, and / or the area of ​​the bounding rectangle corresponding to the point cloud cluster is less than or equal to the area threshold, the point cloud cluster is considered an invalid scattering point, or the point cloud cluster is a ground water stain or a phantom line, or the point cloud cluster was acquired due to sidelobe interference, and therefore the point cloud cluster is discarded.

[0063] In step 105, the attributes of the target point cloud cluster are determined, and the location of the obstacle is determined.

[0064] It should be understood that the attributes of the target point cloud cluster are not limited in the embodiments of this application. In some embodiments, the attributes of the target point cloud cluster include, but are not limited to, at least one of the following: the orientation of the target point cloud cluster, the type of the target point cloud cluster, the position of the target point cloud cluster, the velocity of the target point cloud cluster, and the confidence level of the target point cloud cluster.

[0065] In some embodiments, Figure 4 This application provides a schematic diagram of an implementation process for determining the attributes of a target point cloud cluster, as illustrated in the embodiments of this application. Figure 4 As shown, determining the attributes of the target point cloud cluster further includes the following steps 401 to 403: Step 401: Based on the acquired vehicle status information, predict the vehicle's driving trajectory; Step 402: If the point cloud density of the target point cloud cluster is less than or equal to the second density threshold, the target point cloud cluster is projected onto the driving trajectory; the second density threshold is greater than the first density threshold. Step 403: Determine the tangent direction of the projection point of the target point cloud cluster on the driving trajectory as the orientation of the target point cloud cluster.

[0066] It is understood that in the vehicle control method provided in this application embodiment, the vehicle's driving trajectory is predicted based on the acquired vehicle state information; when the point cloud density of the target point cloud cluster is less than or equal to a second density threshold, the target point cloud cluster is projected onto the driving trajectory; and the tangent direction of the projection point of the target point cloud cluster on the driving trajectory is determined as the orientation of the target point cloud cluster. Thus, when the point cloud density of the target point cloud cluster is lower than a certain threshold, it is considered a missed detection, and it is projected onto the predicted driving trajectory, with the orientation of the target point cloud cluster determined as the tangent direction of the projection point onto the predicted driving trajectory, thereby improving obstacle avoidance rate and driving safety.

[0067] In some embodiments, determining the attributes of the target point cloud cluster includes: if the point cloud density of the target point cloud cluster is greater than a second density threshold, determining the orientation of the target point cloud cluster based on the circumscribed rectangle of the target point cloud cluster.

[0068] It should be understood that, in the embodiments of this application, the point cloud density of the target point cloud cluster is greater than a first density threshold and less than or equal to a second density threshold; the second density threshold is greater than the first density threshold.

[0069] In some embodiments, when the point cloud density of the target point cloud cluster is less than or equal to a second density threshold, projecting the target point cloud cluster onto the driving trajectory includes: determining the centroid of the target point cloud cluster when the point cloud density of the target point cloud cluster is less than or equal to the second density threshold; and projecting the centroid of the target point cloud cluster onto the driving trajectory.

[0070] In some embodiments, the method further includes: determining the attributes of the target point cloud cluster based on the acquired visual image, radar point cloud, and ultrasonic point cloud of the vehicle. Further, in some embodiments, determining the attributes of the target point cloud cluster based on the acquired visual image, radar point cloud, and ultrasonic point cloud of the vehicle includes: extracting features from the visual image, radar point cloud, and ultrasonic point cloud to obtain multimodal features; fusing the multimodal features to obtain fused features; determining the type and attributes of each collected object based on the fused features; determining the target object associated with the target point cloud cluster among the objects; and determining the type and attributes of the target object as the type and attributes of the target point cloud cluster.

[0071] Furthermore, in some embodiments, determining the type and attributes of each collected object based on the fused features includes: inputting the fused features into an object detection (OD) head to obtain the type of each object; and inputting the fused features into an occupancy network (OCC) head to obtain the attributes of each object. The object detection head can output 3D bounding boxes and types, and the occupancy network can output a dense 3D voxel grid and semantic labels.

[0072] In some embodiments, the method further includes: smoothing the updated target point cloud cluster using a Kalman filter algorithm.

[0073] It is understood that in the vehicle control method provided in the embodiments of this application, the Kalman filter algorithm is used to smooth the updated target point cloud cluster, which is beneficial to eliminating the positioning drift caused by beam divergence.

[0074] In step 106, a parking path is generated based on the location of the obstacle to control the vehicle to park based on the parking path.

[0075] The following describes an exemplary application of the embodiments of this application in a real-world application scenario.

[0076] This application relates to the field of autonomous driving environmental perception and multi-sensor data fusion technology. Specifically, it relates to a vision, radar and ultrasonic multi-modal perception fusion method (i.e. an example of a vehicle control method) and system for automatic parking scenarios, based on vehicle kinematic trajectory prediction and geometric contour correction.

[0077] This application provides a multimodal perception fusion method (i.e., an example of a vehicle control method) and system based on trajectory prediction and contour correction to solve the technical problems of redundant calculation of multi-source data fusion, inaccurate ultrasonic contour construction, and misjudgment of target attributes in low-speed parking perception systems.

[0078] In some embodiments, the multimodal perception fusion method based on trajectory prediction and contour correction (i.e., an example of a vehicle control method) includes the following steps S11 to S14: Step S11: Obtain the real-time status information of the vehicle; predict the future driving trajectory of the vehicle based on the vehicle's Ackerman steering geometry model, and generate a dynamic region of interest covering the trajectory. Further, in step S11, generating a dynamic region of interest (i.e., an example of a region of interest) covering the trajectory includes: when the vehicle is in forward gear, generating a forward fan-shaped or rectangular trajectory region of interest along the vehicle's orientation, starting from the center of the front suspension; when the vehicle is in reverse gear, generating a backward trajectory region of interest along the opposite orientation, starting from the center of the rear axle; the width of the dynamic region of interest is dynamically expanded according to the vehicle's physical width and a preset safety distance.

[0079] Step S12: Acquire visual point cloud, radar point cloud and ultrasonic point cloud data; project the visual point cloud, radar point cloud and ultrasonic point cloud into the dynamic ROI area, and perform weighted fusion of the source attributes of the visual point cloud, radar point cloud and ultrasonic point cloud based on the preset source attribute masking rules to remove isolated noise points. Further, in step S12, the weighted fusion of the source attributes of the visual point cloud, radar point cloud, and ultrasonic point cloud based on the preset source attribute masking rules includes: if the target is detected by both the visual sensor and the radar sensor at the same time, and the distance matching error is less than a preset threshold, then a "visual + radar" composite source attribute is assigned, and a bitwise AND operation is performed on the visual point cloud and the radar point cloud; if only ultrasonic detection is performed, the point cloud is retained; and point cloud clusters are determined based on the retained point cloud and the point cloud after the bitwise AND operation.

[0080] Step S13: For point cloud clusters within the ROI region, calculate the bounding rectangle of the minimum area using the rotating caliper method, and combine Kalman filtering to correct the outline and center position of the point cloud clusters. Further, in step S13, the calculation of the circumscribed rectangle with the minimum area using the rotating caliper method includes: obtaining the convex hull (i.e., an example of a convex hull polygon) of the clustered point cloud (i.e., a point cloud cluster) within the ROI region; traversing the edges of the convex hull using the rotating caliper algorithm to solve for the circumscribed rectangle with the minimum area surrounding the point cloud cluster; if the aspect ratio of the circumscribed rectangle exceeds a preset threshold (i.e., an example of an aspect ratio threshold), or the point cloud density within the circumscribed rectangle is lower than a first preset density threshold, then the target point cloud cluster is determined to be an invalid scattering point and is removed, thus obtaining the target point cloud cluster.

[0081] Further, in step S13, the correction of the outline and center position of the point cloud cluster by combining Kalman filtering includes: calculating the centroid of the point cloud cluster and projecting the centroid onto the predicted driving trajectory to find the nearest point (i.e., an example of a projection point); when the point cloud density of the target point cloud cluster is lower than the second preset density threshold, the orientation of the target point cloud cluster is forcibly locked to the tangent direction at the projection point on the predicted driving trajectory, and the target point cloud cluster is smoothly tracked using a Kalman filter.

[0082] Step S14: Based on visual images, radar point clouds, and ultrasonic point clouds, determine the target type, location, velocity, and confidence level of the target point cloud clusters, and other fused target attributes.

[0083] This application also provides a multimodal perception fusion system based on trajectory prediction and contour correction, including: a data receiving module for receiving point cloud data from multiple sensors and vehicle status information; a trajectory prediction module for performing vehicle trajectory prediction and generating dynamic ROI regions; a point cloud processing and contour correction module for performing point cloud filtering and contour correction based on the rotating caliper method; and an attribute fusion module for outputting fusion target attributes according to source attribute masking rules.

[0084] It is understood that the multimodal perception fusion method based on trajectory prediction and contour correction provided in this application can achieve the following technical effects: (1) Significantly improve positioning accuracy: By introducing vehicle kinematics prior to generate dynamic ROI regions, and combining the rotating caliper method to perform geometric contour correction on ultrasonic point cloud clusters and point cloud clusters obtained based on visual point clouds and radar point clouds, the positioning drift caused by beam divergence (e.g., ultrasonic beam divergence) is effectively eliminated. The measured data shows that after the point cloud clusters are corrected, the positioning accuracy of the obstacle center point is improved by 10%; (2) Target attribute accuracy is greatly improved: Through the source attribute masking mechanism and contour density determination, the problem of false obstacle alarms caused by ground water stains, phantom lines and side lobe interference is effectively suppressed, and the attribute misjudgment problem of misjudging the corner of the wall as a stone is solved. The measured data shows that the comprehensive accuracy of target attributes (such as type and orientation) has increased by 100% (i.e., doubled). (3) Reduce system computing load: The dynamic screening mechanism centered on vehicle trajectory avoids dense calculation across the entire field of view, greatly reducing the load on the central processing unit (CPU) of the embedded controller and meeting the real-time requirements of the automatic parking system.

[0085] Figure 5 A schematic diagram of the implementation process of a vehicle control method provided in this application embodiment. Figure 2 ,like Figure 5 As shown, the vehicle control method includes the following steps S21 to S25: S21, Obtain vehicle status information; S22, predicts trajectories and generates dynamic ROIs based on the Ackerman model; S23, Multi-source point cloud (visual, radar, ultrasonic) filtering and denoising; S24, Contour correction based on the rotating caliper method; S25, source attribute mask weighted fusion.

[0086] Figure 6A schematic diagram of the implementation process of a ROI region provided in this application embodiment. Figure 1 ,like Figure 6 As shown, 601 represents the vehicle, 602 represents the vehicle's historical trajectory, 603 represents the predicted vehicle's driving trajectory, 604 represents the ROI region, 605 represents the point cloud data, 606 represents the output target, 607 represents the ROI region corresponding to the visual image, and 608 represents the bounding rectangle of a point cloud cluster; the visual image is used to provide attributes for the point cloud data, and the front axis coordinates of vehicle 601 are (-2.239522, 0.838265).

[0087] Figure 7 A schematic diagram of the implementation process of a ROI region provided in this application embodiment. Figure 2 ,like Figure 7 As shown, 701 represents the vehicle, 702 represents the vehicle's historical trajectory, 703 represents the predicted vehicle's driving trajectory, 704 represents the ROI region, 705 represents the point cloud data, 706 represents the output target, and 707 represents the ROI region corresponding to the visual image, which is used to provide attributes for the point cloud data. The front axis coordinates of vehicle 601 are (-0.048665, 2.269848).

[0088] In some embodiments, this application provides an efficient multimodal perception fusion method for automatic parking (i.e., an example of a vehicle control method), which includes the following steps S31 to S34: Step S31: Dynamic Trajectory and ROI Region Generation. The vehicle's gear position (D / R), front wheel steering angle, and vehicle speed are acquired in real time. The vehicle's turning radius is calculated based on the Ackerman steering geometry model. The vehicle's trajectory is predicted based on the turning radius; the vehicle's ROI region is generated based on the turning radius and gear position.

[0089] If the vehicle is in drive, a forward fan-shaped ROI area is generated starting from the front overhang of the vehicle; if it is in reverse, a backward fan-shaped ROI area is generated starting from the rear axle of the vehicle. The width of the ROI area dynamically expands with the vehicle width and safety margin, forming a "dynamic sensing funnel".

[0090] Step S32: Spatiotemporal Alignment and Filtering of Multi-Source Point Clouds. The point cloud data from the surround-view camera, millimeter-wave radar, and ultrasonic radar are uniformly transformed to the vehicle coordinate system (BaseLink). Using a pose transformation matrix, the radar point cloud, visual point cloud, and ultrasonic point cloud are projected onto the dynamic ROI region generated in step S31. Simultaneously, using the visual point cloud, radar point cloud, and ultrasonic point cloud, morphological operations (such as dilation and erosion) and bitwise AND operations are performed to remove isolated noise and ground reflection interference, resulting in a superimposed point cloud cluster.

[0091] In step S32, the bitwise AND operation includes: source attribute mask fusion, defining multiple source attribute masks (such as VisionAVM, RadarFront, USS_RAW). All targets are traversed; if a target is detected by both the visual sensor and the radar sensor simultaneously, and the distance matching error is less than 1.8 meters, then the visual point cloud and the radar point cloud are subjected to a bitwise AND operation to obtain a superimposed point cloud; if the target is only detected by ultrasonic waves, the point cloud is retained.

[0092] Step S33: Contour Correction Based on Rotating Calipers. For the ultrasonic clustered point cloud (i.e., an example of an ultrasonic point cloud cluster) within the ROI region and the point cloud cluster obtained from the superimposed point cloud, its centroid is first calculated. Then, the convex hull of the point cloud cluster is calculated, and the Rotating Calipers method is applied to solve for the minimum area circumscribed rectangle. This corrects the defect of ultrasonic waves having "distance but no angle." If the area of ​​the calculated circumscribed rectangle is too small or the point cloud density is below the first density threshold (e.g., area parameter less than 5), it is determined to be invalid scattering and discarded. If the point cloud density of the cluster is greater than the first density threshold and less than or equal to the second density threshold, the orientation of the point cloud cluster is forcibly locked to the tangent direction of the projection point of the centroid of the cluster on the predicted driving trajectory, and a simplified Kalman filter is used to smooth the target point cloud cluster, eliminating inter-frame jitter.

[0093] Step S34: Based on visual images, radar point clouds, and ultrasonic point clouds, assign cluster attributes to the target point cloud and finally output a fused target list containing precise location, velocity, type, and source attributes for use by the downstream planning and control module.

[0094] Through the above implementation methods, the efficient multimodal perception fusion method for automatic parking provided in this application can make up for the deficiencies of physical sensors by introducing geometric constraints and prior knowledge at the algorithm level without increasing hardware costs, and achieve a double leap in positioning accuracy and attribute accuracy.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of this application without departing from the spirit and scope of the technical solutions of this application, and all such modifications or substitutions should be covered within the protection scope of this application.

[0096] It should be noted that although the steps of the method in this application are described in a specific order in the accompanying drawings, this does not require or imply that these steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps; or steps from different embodiments may be combined into a new technical solution. Based on the foregoing embodiments, this application provides an apparatus comprising the included modules and the units included in each module, which can be implemented by a processor; of course, it can also be implemented by specific logic circuits; in implementation, the processor can be an AI acceleration engine (such as an NPU), a graphics processing unit (GPU), a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field-programmable gate array (FPGA), etc.

[0097] Figure 8 This is a schematic diagram of a vehicle control device provided in an embodiment of this application, such as... Figure 8 As shown, the vehicle control device 80 includes: a first determining module 801, a second determining module 802, a third determining module 803, a fourth determining module 804, a fifth determining module 805, and a control module 806; wherein, The first determining module 801 is configured to determine the ROI area of ​​the vehicle in response to the vehicle entering the parking mode; The second determining module 802 is configured to determine the point cloud clusters corresponding to the sensing data located in the ROI region; the sensing data includes: visual point cloud, radar point cloud and ultrasonic point cloud. The third determining module 803 is configured to determine the bounding rectangle of each point cloud cluster. The fourth determining module 804 is configured to determine the target point cloud cluster from the point cloud cluster based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster. The fifth determining module 805 is configured to determine the attributes of the target point cloud cluster and determine the location of the obstacle; The control module 806 is configured to generate a parking path based on the location of the obstacle, so as to control the vehicle to park based on the parking path.

[0098] In some embodiments, determining a target point cloud cluster from the point cloud cluster based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster includes at least one of the following: removing the point cloud cluster if the point cloud density of the point cloud cluster is less than or equal to a first density threshold; removing the point cloud cluster if the aspect ratio of the bounding rectangle corresponding to the point cloud cluster is greater than or equal to an aspect ratio threshold; and removing the point cloud cluster if the area of ​​the bounding rectangle corresponding to the point cloud cluster is less than or equal to an area threshold.

[0099] In some embodiments, the second determining module 802 is further configured to perform dilation and / or erosion operations on the visual point cloud, radar point cloud, and ultrasonic point cloud located in the ROI region, respectively, to obtain processed visual point cloud, processed radar point cloud, and processed ultrasonic point cloud; to superimpose the processed radar point cloud and the processed visual point cloud to obtain a superimposed point cloud; and to cluster the processed ultrasonic point cloud and the superimposed point cloud to obtain multiple point cloud clusters.

[0100] In some embodiments, the fifth determining module 805 is further configured to predict the driving trajectory of the vehicle based on the acquired state information of the vehicle; if the point cloud density of the target point cloud cluster is less than or equal to a second density threshold, project the target point cloud cluster onto the driving trajectory; the second density threshold is greater than a first density threshold; and determine the tangent direction of the projection point of the target point cloud cluster on the driving trajectory as the orientation of the target point cloud cluster.

[0101] In some embodiments, the third determining module 803 is further configured to determine the convex hull polygon of the point cloud cluster; using a rotating caliper algorithm, traversing each edge of the convex hull polygon, constructing a corresponding circumscribed rectangle for each edge, and determining the area of ​​each circumscribed rectangle; and determining the circumscribed rectangle with the smallest area as the circumscribed rectangle of the point cloud cluster.

[0102] In some embodiments, the vehicle control device 80 further includes a smoothing module; the smoothing module is configured to smooth the updated target point cloud cluster using a Kalman filter algorithm.

[0103] The descriptions of the above device embodiments are similar to those of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0104] It should be noted that the module division in the embodiments of this application is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, exist as separate physical units, or have two or more units integrated into one unit. The integrated units can be implemented in hardware, as software functional units, or a combination of software and hardware.

[0105] It should be noted that, in the embodiments of this application, if the above methods are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause the vehicle to execute all or part of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0106] This application provides a vehicle, Figure 9 This application provides a schematic diagram of the structure of a vehicle, as shown in the embodiment of the present application. Figure 9 As shown, the vehicle 90 includes a memory 901 and a processor 902. The memory 901 stores a computer program that can run on the processor 902. When the processor 902 executes the program, it implements the steps in the method provided in the above embodiments.

[0107] It should be noted that the memory 901 is configured to store instructions and applications executable by the processor 902, and can also cache data to be processed or already processed in the processor 902 and various modules in the vehicle 90 (e.g., image data, audio data, voice communication data and video communication data), which can be implemented by flash memory or random access memory (RAM).

[0108] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method provided in the above embodiments.

[0109] This application provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the steps in the method provided in the above-described method embodiments.

[0110] It should be noted that the descriptions of the storage medium and vehicle embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and vehicle embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0111] It should be understood that the phrases "one embodiment," "an embodiment," or "some embodiments" mentioned throughout the specification mean that a specific feature, structure, or characteristic related to an embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment," "in one embodiment," or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. The descriptions of the various embodiments above tend to emphasize the differences between the various embodiments; their similarities or commonalities can be referred to mutually, and for the sake of brevity, they will not be repeated here.

[0112] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three kinds of relationships. For example, object A and / or object B can represent three situations: object A exists alone, object A and object B exist simultaneously, and object B exists alone.

[0113] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or vehicle. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or vehicle that includes that element.

[0114] In the several embodiments provided in this application, it should be understood that the disclosed vehicles and methods can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple modules or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between vehicles or modules can be electrical, mechanical, or other forms.

[0115] The modules described above as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules. They may be located in one place or distributed across multiple network units. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0116] In addition, each functional module in the various embodiments of this application can be integrated into one processing unit, or each module can be a separate unit, or two or more modules can be integrated into one unit; the integrated modules can be implemented in hardware or in the form of hardware plus software functional units.

[0117] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0118] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium.

[0119] Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause the vehicle to execute all or part of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0120] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined to obtain new method embodiments without conflict. The features disclosed in the several product embodiments provided in this application can be arbitrarily combined to obtain new product embodiments without conflict. The features disclosed in the several method or vehicle embodiments provided in this application can be arbitrarily combined to obtain new method embodiments or vehicle embodiments without conflict.

[0121] The above are merely embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A vehicle control method, characterized in that, The method includes: In response to the vehicle entering parking mode, the ROI area of ​​the vehicle is determined; Identify the point cloud clusters corresponding to the sensing data located in the ROI region; the sensing data includes: visual point clouds, radar point clouds, and ultrasonic point clouds. For each of the aforementioned point cloud clusters, determine the bounding rectangle of the point cloud cluster; Based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster, the target point cloud cluster is determined from the point cloud cluster. Determine the attributes of the target point cloud cluster and the location of obstacles; A parking path is generated based on the location of the obstacle to control the vehicle to park based on the parking path.

2. The vehicle control method according to claim 1, characterized in that, Determining the target point cloud cluster from the point cloud cluster based on the point cloud density and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster includes at least one of the following: If the point cloud density of the point cloud cluster is less than or equal to a first density threshold, the point cloud cluster is discarded. If the aspect ratio of the bounding rectangle corresponding to the point cloud cluster is greater than or equal to the aspect ratio threshold, the point cloud cluster is discarded. If the area of ​​the bounding rectangle corresponding to the point cloud cluster is less than or equal to the area threshold, the point cloud cluster is discarded.

3. The vehicle control method according to claim 1, characterized in that, The step of determining the point cloud clusters corresponding to the sensing data located in the ROI region includes: Dilation and / or erosion operations are performed on the visual point cloud, radar point cloud and ultrasonic point cloud located in the ROI region to obtain the processed visual point cloud, processed radar point cloud and processed ultrasonic point cloud respectively. The processed radar point cloud and the processed visual point cloud are superimposed to obtain a superimposed point cloud. The processed ultrasonic point cloud and the superimposed point cloud are clustered to obtain multiple point cloud clusters.

4. The vehicle control method according to any one of claims 1 to 3, characterized in that, Determining the attributes of the target point cloud cluster further includes: Based on the acquired vehicle status information, predict the vehicle's driving trajectory; If the point cloud density of the target point cloud cluster is less than or equal to the second density threshold, the target point cloud cluster is projected onto the driving trajectory; the second density threshold is greater than the first density threshold. The tangent direction of the projection point of the target point cloud cluster on the driving trajectory is determined as the orientation of the target point cloud cluster.

5. The vehicle control method according to any one of claims 1 to 3, characterized in that, Determining the circumscribed rectangle of the point cloud cluster includes: Determine the convex hull polygon of the point cloud cluster; Using the rotating caliper algorithm, traverse each edge of the convex hull polygon, construct a corresponding circumscribed rectangle for each edge, and determine the area of ​​each circumscribed rectangle; The outer rectangle with the smallest area is determined as the outer rectangle of the point cloud cluster.

6. The vehicle control method according to claim 4, characterized in that, The method further includes: The updated target point cloud cluster is smoothed using the Kalman filter algorithm.

7. A vehicle control device, characterized in that, The device includes: The first determining module is configured to determine the ROI area of ​​the vehicle in response to the vehicle entering parking mode; The second determining module is configured to determine the point cloud clusters corresponding to the sensing data located in the ROI region; the sensing data includes: visual point cloud, radar point cloud and ultrasonic point cloud. The third determining module is configured to determine the bounding rectangle of each point cloud cluster. The fourth determining module is configured to determine the target point cloud cluster from the point cloud cluster based on the point cloud density of the point cloud cluster and / or the geometric features of the bounding rectangle corresponding to the point cloud cluster. The fifth determining module is configured to determine the attributes of the target point cloud cluster and determine the location of obstacles; The control module is configured to generate a parking path based on the location of the obstacle, so as to control the vehicle to park based on the parking path.

8. A vehicle, characterized in that, The vehicle includes a memory and a processor; the memory stores a computer program that can run on the processor, and the processor executes the program to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the method described in any one of claims 1 to 6.

Images