CN-121979201-A - Image recognition-based tour electric vehicle dynamic obstacle avoidance method and system

CN121979201ACN 121979201 ACN121979201 ACN 121979201ACN-121979201-A

Abstract

The application discloses a method and a system for dynamically avoiding obstacles of an electric tour automobile based on image recognition, and relates to the technical field of automobile obstacle avoidance, wherein the method comprises the steps of obtaining tour characteristic information of a target tour automobile and establishing a standard tour management and control area; the method comprises the steps of collecting multi-angle environment images, carrying out object identification on the obtained multi-angle environment images, carrying out follow-up correction on a standard tour management and control area by combining a real-time coordinate set and corresponding real-time tour action time sequence information, obtaining a real-time tour management and control body, carrying out collision risk judgment, and controlling a target tour vehicle to carry out dynamic obstacle avoidance according to a collision risk judgment result. The technical problems that the conventional electric automobile has single obstacle avoidance strategy, and the normal tour efficiency is influenced by response lag or excessive obstacle avoidance are solved.

Inventors

WANG HAONAN
LIANG HONGLIANG
ZHAO DONGMING

Assignees

北京星凯达舞台技术有限公司

Dates

Publication Date: 20260505
Application Date: 20251225

Claims (10)

1. The image recognition-based tour electric vehicle dynamic obstacle avoidance method is characterized by comprising the following steps of: Obtaining tour characteristic information of a target tour vehicle, and establishing a standard tour management and control area according to the tour characteristic information; Acquiring a multi-angle environment image through a distributed image acquisition device, and carrying out object recognition on the acquired multi-angle environment image to acquire a real-time coordinate set, wherein the real-time coordinate set comprises a self-vehicle correction coordinate and a list object coordinate; The real-time coordinate set and the corresponding real-time tour action time sequence information are combined, the standard tour control area is corrected in a follow-up mode, a real-time tour control body is obtained, and collision risk judgment is carried out according to the real-time tour control body; And controlling the target tour vehicle to dynamically avoid the obstacle according to the collision risk judging result.
2. The method for dynamically avoiding the obstacle for the tour electric vehicle based on the image recognition as set forth in claim 1, wherein the step of obtaining tour characteristic information of the target tour vehicle comprises the steps of: Obtaining tour route information of a target environment, wherein the tour route information is provided with a first time sequence mark; Acquiring tour motion information of a target tour vehicle, wherein the tour motion information comprises time sequence motion information of at least one movable component on the target tour vehicle, and a time stamp of the time sequence motion information forms a second time sequence mark; And according to the first time sequence mark and the second time sequence mark, performing time sequence alignment on the tour route information and the tour action information to acquire the tour characteristic information.
3. The image recognition-based tour electric vehicle dynamic obstacle avoidance method as set forth in claim 2, wherein establishing a standard tour management and control area according to the tour characteristic information includes: acquiring intrinsic design data of a target tour vehicle, traversing and establishing a digital space model of the target tour vehicle and a plurality of movable components according to the intrinsic design data, and acquiring a block model set; Envelope analysis is carried out by combining the tour motion information and the block model set to generate a block-envelope model set, wherein the block-envelope model set comprises a base space model, a movable space model and an envelope space model; Establishing a dynamic contour model of the target tour vehicle based on the block-envelope model set, wherein the dynamic contour model has a time sequence mark; and integrating the tour route information and the dynamic contour model to perform tour evolution based on a time sequence mark, and generating the standard tour management and control area.
4. The image recognition-based tour electric vehicle dynamic obstacle avoidance method as set forth in claim 1, wherein the steps of performing multi-angle environmental image acquisition by a distributed image acquisition device, performing object recognition on the acquired multi-angle environmental image, and acquiring a real-time coordinate set include: acquiring the multi-angle environment image through image acquisition devices arranged on the vehicle side and the road side; preprocessing the multi-angle environment image, and removing background objects in a preset exclusion list; Performing dual object recognition on the preprocessed multi-angle environment image by combining a computer vision method, wherein the dual object recognition comprises the steps of recognizing vehicle body characteristics to obtain the self-vehicle correction coordinates, and recognizing specified objects in a preset list to obtain list object coordinates; And acquiring real-time tour motion time sequence information of a movable part on the target tour vehicle in real time through a vehicle-mounted sensor, and outputting the self-vehicle correction coordinates, the list object coordinates and the real-time tour motion time sequence information together into the real-time coordinate set.
5. The image recognition-based tour electric vehicle dynamic obstacle avoidance method as set forth in claim 1, wherein the multi-angle environmental image is acquired by a distributed image acquisition device, and the acquired multi-angle environmental image is subject to recognition, and a real-time coordinate set is acquired, further comprising: According to the acquisition time stamp of the list object coordinates, matching and acquiring list object preamble coordinates of the list object; And combining the list object coordinates with the list object preamble coordinates, calculating and determining object motion vector information, and storing the object motion vector information and the list object coordinates in the real-time coordinate set in an associated mode.
6. The method for dynamically avoiding the obstacle for the tour electric automobile based on the image recognition according to claim 1, wherein the method for carrying out the follow-up correction on the standard tour management and control area by combining the real-time coordinate set and the corresponding real-time tour action time sequence information, obtaining a real-time tour management and control body, and carrying out collision risk discrimination according to the real-time tour management and control body comprises the following steps: Defining the time length of a follow-up correction window based on the total response time of the system; the total response time of the system is the sum of perceived processing delay, vehicle braking response time and preset safety redundancy time which are determined based on statistical analysis.
7. The image recognition-based tour electric vehicle dynamic obstacle avoidance method as set forth in claim 1, wherein the performing a follow-up correction on the standard tour management and control area by combining the real-time coordinate set and the corresponding real-time tour action time sequence information to obtain a real-time tour management and control body includes: Traversing the quasi-tour management and control area according to the real-time tour action time sequence information to determine a real-time dynamic contour model of the target tour vehicle; Performing interval extraction on the standard tour-control area by combining a predefined follow-up correction window and the real-time dynamic profile model, and generating a standard tour-control body by combining a preset clearance constraint and an interval extraction result; And carrying out follow-up correction on the standard tour management and control body according to the self-vehicle correction coordinates, wherein the follow-up correction comprises position correction and route correction, and the real-time tour management and control body is generated.
8. The image recognition-based tour electric vehicle dynamic obstacle avoidance method as set forth in claim 1, wherein the collision risk discrimination is performed according to the real-time tour management and control body, comprising: acquiring object motion vector information corresponding to the list object coordinates; combining the list object coordinates, the object motion vector information and a predefined follow-up correction window to construct predicted time sequence track information of the list object; performing space-time intersection operation on the basis of the predicted time sequence track information and the real-time tour management and control body; If the space-time intersection operation result is that an intersection exists, correspondingly calculating collision reaction time, determining a collision risk level output as a collision risk judging result according to the collision reaction time, the object motion vector information and a preset collision risk judging matrix, and otherwise, outputting a collision risk judging result as null.
9. The image recognition-based tour electric vehicle dynamic obstacle avoidance method as set forth in claim 1, wherein the step of controlling the target tour vehicle to dynamically avoid the obstacle according to the collision risk discrimination result comprises the steps of: if the collision risk judging result is not null, controlling the target tour vehicle to dynamically avoid the obstacle according to a grading obstacle avoidance strategy, wherein the grading obstacle avoidance strategy control comprises the following steps: If the collision risk judging result corresponds to a preset low risk category, sending an early warning signal and controlling the target tour vehicle to slightly slow down; If the collision risk judging result corresponds to a preset risk range, sending an early warning signal, controlling the target tour vehicle to slightly slow down, and stopping tour actions of movable components on the target tour vehicle for entropy increase of the standard tour control area; and if the collision risk judging result corresponds to a preset high risk category, sending out an early warning signal and controlling the target tour vehicle to brake and avoid the obstacle.
10. The image recognition-based tour electric vehicle dynamic obstacle avoidance system, which is characterized by being used for executing the image recognition-based tour electric vehicle dynamic obstacle avoidance method as claimed in any one of claims 1-9, comprising: The information acquisition module is used for acquiring tour characteristic information of the target tour vehicle and establishing a standard tour management and control area according to the tour characteristic information; The image acquisition module is used for acquiring multi-angle environment images through the distributed image acquisition device, and carrying out object recognition on the acquired multi-angle environment images to acquire a real-time coordinate set, wherein the real-time coordinate set comprises self-vehicle correction coordinates and list object coordinates; the area correction module is used for carrying out follow-up correction on the standard tour control area by combining the real-time coordinate set and the corresponding real-time tour action time sequence information, acquiring a real-time tour control body and judging collision risk according to the real-time tour control body; and the vehicle obstacle avoidance module is used for controlling the target tour vehicle to dynamically avoid the obstacle according to the collision risk discrimination result.

Description

Image recognition-based tour electric vehicle dynamic obstacle avoidance method and system Technical Field The application relates to the technical field of automobile obstacle avoidance, in particular to a tour electric automobile dynamic obstacle avoidance method and system based on image recognition. Background With the acceleration of the urban process and the development of sharing economy, the tour electric automobile is widely applied to the scenes of urban short-distance connection, scenic spot tour, park commute and the like with the characteristics of environmental protection and flexibility. Tour electric vehicles are usually operated in areas with complex environments, and the dynamic obstacle avoidance capability of the tour electric vehicles is directly related to operation safety and service quality. In the prior art, the traditional automobile obstacle avoidance method is used for acquiring image information only through a single camera to cause a dead zone in environmental perception, and meanwhile, when a safety control area is built, a static or simple dynamic bounding box model is mostly adopted, so that the automatic inspection vehicle cannot be fully combined with the self-inspection characteristics of the inspection vehicle and the real-time motion state of surrounding objects to dynamically adjust, the accuracy of collision risk judgment is lower, the requirement of the inspection electric vehicle on fine obstacle avoidance under a complex dynamic environment is difficult to meet, and the problems that the obstacle avoidance strategy is single, response is lag or excessive obstacle avoidance influences normal inspection efficiency and the like exist. Disclosure of Invention The embodiment of the application solves the technical problems that the conventional electric automobile has single obstacle avoidance strategy, delayed response or excessive obstacle avoidance influences the normal tour efficiency by providing the tour electric automobile dynamic obstacle avoidance method and system based on image recognition. The technical scheme for solving the technical problems is as follows: In a first aspect, the application provides a tour electric vehicle dynamic obstacle avoidance method based on image recognition, which comprises the following steps: Obtaining tour characteristic information of a target tour vehicle, and establishing a standard tour management and control area according to the tour characteristic information; Acquiring a multi-angle environment image through a distributed image acquisition device, and carrying out object recognition on the acquired multi-angle environment image to acquire a real-time coordinate set, wherein the real-time coordinate set comprises a self-vehicle correction coordinate and a list object coordinate; The real-time coordinate set and the corresponding real-time tour action time sequence information are combined, the standard tour control area is corrected in a follow-up mode, a real-time tour control body is obtained, and collision risk judgment is carried out according to the real-time tour control body; And controlling the target tour vehicle to dynamically avoid the obstacle according to the collision risk judging result. In a second aspect, the present application provides an image recognition-based tour electric vehicle dynamic obstacle avoidance system, including: The information acquisition module is used for acquiring tour characteristic information of the target tour vehicle and establishing a standard tour management and control area according to the tour characteristic information; The image acquisition module is used for acquiring multi-angle environment images through the distributed image acquisition device, and carrying out object recognition on the acquired multi-angle environment images to acquire a real-time coordinate set, wherein the real-time coordinate set comprises self-vehicle correction coordinates and list object coordinates; the area correction module is used for carrying out follow-up correction on the standard tour control area by combining the real-time coordinate set and the corresponding real-time tour action time sequence information, acquiring a real-time tour control body and judging collision risk according to the real-time tour control body; and the vehicle obstacle avoidance module is used for controlling the target tour vehicle to dynamically avoid the obstacle according to the collision risk discrimination result. The application provides one or more technical schemes, which at least have the following technical effects or advantages: The embodiment of the application provides a method and a system for dynamically avoiding obstacles of an electric vehicle in tour based on image recognition, which are characterized in that firstly, tour route information and tour motion information of a target tour vehicle are obtained, and the running track of the vehicle at different time nodes and the motion rules of