CN-121386883-B - Multi-view vision obstacle avoidance method and system using image sensor

CN121386883BCN 121386883 BCN121386883 BCN 121386883BCN-121386883-B

Abstract

The invention is suitable for the technical field of multi-view visual obstacle avoidance, and provides a multi-view visual obstacle avoidance method and system using an image sensor. The method comprises the steps of performing environment multi-view monitoring shooting, determining target barriers, recording relevant position data, planning an obstacle avoidance selection area, selecting a plurality of obstacle avoidance positions, calculating the expected obstacle avoidance values of the obstacle avoidance positions, judging whether the obstacle avoidance conditions are met, determining final obstacle avoidance positions when the obstacle avoidance conditions are met, performing obstacle avoidance movement control, determining an enlarged selection area when the obstacle avoidance conditions are not met, and reselecting the plurality of obstacle avoidance positions. The method has the advantages that the obstacle avoidance selection area can be planned, the plurality of obstacle avoidance positions are selected, the corresponding obstacle avoidance expected value is calculated, whether the obstacle avoidance condition is met is judged in a comparison mode, when the obstacle avoidance condition is not met, the selection area is enlarged, the plurality of obstacle avoidance positions are reselected, the obstacle avoidance can be realized, meanwhile, the selected final obstacle avoidance position is enabled to have small influence on the whole transportation process.

Inventors

YUAN GANG

Assignees

湖南金康光电有限公司

Dates

Publication Date: 20260508
Application Date: 20251202

Claims (8)

1. The multi-view visual obstacle avoidance method using the image sensor is characterized by comprising the following steps of: in the moving process, performing multi-view monitoring shooting of the environment, acquiring multi-view monitoring data, identifying the multi-view monitoring data, determining a target obstacle, and recording relevant position data; Performing obstacle avoidance analysis on the related position data, planning an obstacle avoidance selection area, and selecting a plurality of obstacle avoidance positions from the obstacle avoidance selection area; calculating the expected obstacle avoidance values of a plurality of obstacle-avoidance positions based on the related position data, comparing the expected obstacle-avoidance values with a preset threshold value, and judging whether an obstacle-avoidance condition is met; When the obstacle avoidance conditions are met, determining a final obstacle avoidance position from a plurality of obstacle-avoidance positions, and performing obstacle avoidance movement control; when the obstacle avoidance conditions are not met, expanding and planning the obstacle avoidance selection area, determining an expanded selection area, and reselecting a plurality of obstacle avoidance positions; Based on the related position data, calculating expected obstacle avoidance values of a plurality of obstacle-avoidance positions, comparing the expected obstacle avoidance values with a preset threshold, and judging whether the obstacle avoidance conditions are met or not specifically comprises the following steps: calculating obstacle distances and terminal distances corresponding to a plurality of obstacle-simulated positions based on the related position data; according to the obstacle distances and the end point distances, carrying out obstacle avoidance gain analysis of potential energy influence, and calculating the expected obstacle avoidance values of the obstacle-planning positions; comparing a plurality of expected values of the obstacle avoidance with each other, and selecting the maximum expected value; Comparing the maximum expected value with a preset threshold; When the maximum expected value is greater than a preset threshold value, judging that the obstacle avoidance condition is met; When the maximum expected value is smaller than or equal to a preset threshold value, judging that the obstacle avoidance condition is not met; the calculation formulas of the expected value of the obstacle avoidance of the plurality of obstacle-avoidance positions are as follows: ; ; Wherein, the Represents the first The expected value of the obstacle avoidance at the individual obstacle-avoidance locations, For a preset instant prize to be awarded, Is a preset discount factor that is used to determine the discount, Is the maximum expected value that is preset, For the preset suction gain, the suction gain is set, Represents the first The end point distance of each obstacle-avoiding position, For a preset repulsive gain, Represents the first The obstacle distance of each obstacle-like position, The maximum influence distance for the preset obstacle.
2. The method for avoiding obstacle with multiple vision using image sensor according to claim 1, wherein the steps of performing multiple monitoring shots of the environment during the moving process, obtaining multiple monitoring data, identifying the multiple monitoring data, determining the target obstacle, and recording the relevant position data specifically include the following steps: In the moving process, carrying out multi-view monitoring shooting of the environment to obtain multi-view monitoring data; importing obstacle characteristic data; identifying the multi-view monitoring data based on the obstacle characteristic data, and determining a target obstacle; And carrying out positioning analysis on the multi-view monitoring data, and recording relevant position data.
3. The method for avoiding obstacle in view of multiple views using image sensors according to claim 2, wherein said positioning analysis of said multiple views of the monitored data, recording the relevant position data, comprises the steps of: Performing obstacle positioning analysis on the multi-view monitoring data to determine the obstacle distance and the obstacle direction of the target obstacle; Importing positioning characteristic data; Identifying the multi-view monitoring data based on the positioning feature data, determining a plurality of positioning feature entities and determining a plurality of entity distances; acquiring entity positions of a plurality of positioning feature entities; According to the entity positions and the entity distances, current positioning analysis is carried out, and the current positioning position is determined; Determining an obstacle position of the target obstacle according to the current positioning position, the obstacle distance and the obstacle direction; Acquiring an end position; And sorting the current positioning position, the obstacle position and the end position, and recording related position data.
4. A multi-view obstacle avoidance method using image sensors as claimed in claim 3 wherein said performing an obstacle avoidance analysis on said correlated positional data, planning an obstacle avoidance selection area, and selecting a plurality of obstacle-avoidance locations from said obstacle avoidance selection area comprises the steps of: Determining an initial selection distance; planning an obstacle avoidance selection area according to the current positioning position and the initial selection distance; acquiring an obstacle avoidance area map of the obstacle avoidance selection area; and selecting a plurality of obstacle-avoidance positions based on the obstacle-avoidance area map.
5. A method of multi-view obstacle avoidance using an image sensor as claimed in claim 3 wherein said determining a final obstacle avoidance location from a plurality of said obstacle avoidance locations and performing obstacle avoidance movement control comprises the steps of: determining a final obstacle avoidance position corresponding to the maximum expected value from a plurality of obstacle avoidance positions; planning an obstacle avoidance movement route according to the current positioning position and the final obstacle avoidance position; And carrying out obstacle avoidance movement control according to the obstacle avoidance movement route.
6. The method for avoiding obstacle with multiple views using image sensors as recited in claim 4, wherein the expanding the obstacle avoidance area, determining the expanded area, and reselecting the plurality of obstacle-avoidance locations comprises the steps of: Determining an expanded length; Determining an expansion distance according to the expansion length and the initial selection distance; Determining an expansion area according to the current positioning position and the expansion distance; Hiding the obstacle avoidance selection area from the expansion area to obtain an expansion selection area; acquiring an enlarged area map of the enlarged selection area; and reselecting a plurality of obstacle-avoidance locations based on the enlarged area map.
7. The multi-view vision obstacle avoidance system using the image sensor is characterized by comprising a multi-view monitoring shooting module, an area planning processing module, an expected calculation comparison module, an obstacle avoidance movement control module and an area expansion planning module, wherein: The multi-view monitoring shooting module is used for carrying out multi-view monitoring shooting of the environment in the moving process, acquiring multi-view monitoring data, identifying the multi-view monitoring data, determining a target obstacle and recording related position data; The area planning processing module is used for carrying out obstacle avoidance analysis on the relevant position data, planning an obstacle avoidance selection area and selecting a plurality of obstacle avoidance positions from the obstacle avoidance selection area; the expected calculation comparison module is used for calculating expected obstacle avoidance values of a plurality of obstacle avoidance positions based on the relevant position data, comparing the expected obstacle avoidance values with a preset threshold value and judging whether an obstacle avoidance condition is met; the obstacle avoidance movement control module is used for determining a final obstacle avoidance position from a plurality of obstacle-avoidance-simulated positions and performing obstacle avoidance movement control when the obstacle avoidance conditions are met; The area expansion planning module is used for carrying out expansion planning on the obstacle avoidance selection area when the obstacle avoidance condition is not met, determining an expansion selection area and reselecting a plurality of obstacle-avoidance positions; Based on the related position data, calculating expected obstacle avoidance values of a plurality of obstacle-avoidance positions, comparing the expected obstacle avoidance values with a preset threshold, and judging whether the obstacle avoidance conditions are met specifically comprises: calculating obstacle distances and terminal distances corresponding to a plurality of obstacle-simulated positions based on the related position data; according to the obstacle distances and the end point distances, carrying out obstacle avoidance gain analysis of potential energy influence, and calculating the expected obstacle avoidance values of the obstacle-planning positions; comparing a plurality of expected values of the obstacle avoidance with each other, and selecting the maximum expected value; Comparing the maximum expected value with a preset threshold; When the maximum expected value is greater than a preset threshold value, judging that the obstacle avoidance condition is met; When the maximum expected value is smaller than or equal to a preset threshold value, judging that the obstacle avoidance condition is not met; the calculation formulas of the expected value of the obstacle avoidance of the plurality of obstacle-avoidance positions are as follows: ; ; Wherein, the Represents the first The expected value of the obstacle avoidance at the individual obstacle-avoidance locations, For a preset instant prize to be awarded, Is a preset discount factor that is used to determine the discount, Is the maximum expected value that is preset, For the preset suction gain, the suction gain is set, Represents the first The end point distance of each obstacle-avoiding position, For a preset repulsive gain, Represents the first The obstacle distance of each obstacle-like position, The maximum influence distance for the preset obstacle.
8. The system of claim 7, wherein the multi-view monitoring camera module specifically comprises: the image sensor is used for carrying out multi-view monitoring shooting of the environment in the moving process to obtain multi-view monitoring data; A data importing unit for importing obstacle characteristic data; An obstacle recognition unit, configured to recognize the multiple monitoring data based on the obstacle feature data, and determine a target obstacle; and the positioning analysis unit is used for performing positioning analysis on the multi-view monitoring data and recording related position data.

Description

Multi-view vision obstacle avoidance method and system using image sensor Technical Field The invention belongs to the technical field of multi-view visual obstacle avoidance, and particularly relates to a multi-view visual obstacle avoidance method and system using an image sensor. Background The multi-view visual obstacle avoidance technology is characterized in that three-dimensional perception and spatial modeling are carried out on the surrounding environment through two or more cameras, so that the obstacle avoidance technology of automatic detection, distance measurement and path avoidance of the obstacle is realized, image data under different visual angles can be obtained, and a depth map or a point cloud model of the environment is generated by utilizing visual algorithms such as image matching, three-dimensional reconstruction, depth estimation and the like so as to identify the position, shape and motion state of the obstacle. Compared with monocular vision, the multi-eye vision obstacle avoidance device has the advantages of high depth perception precision, strong environmental adaptability, comprehensive space understanding capability and the like, can accurately identify and avoid obstacles in a complex environment, is widely applied to the fields of unmanned aerial vehicle, unmanned aerial vehicle flight, robot navigation and the like, and is one of core technologies for realizing high-precision environment perception and autonomous decision. Among the multi-eye visual obstacle avoidance, the binocular visual obstacle avoidance has the most wide application. In the prior art, binocular vision obstacle avoidance mainly focuses on accurate recognition and distance measurement of an obstacle to realize detection and positioning of the obstacle, however, for selection of the obstacle avoidance position, a specific technical scheme is lacking, and usually only after recognition and positioning of the obstacle are completed, a position capable of avoiding the obstacle is temporarily selected to bypass through simple geometric deviation or local path adjustment, and the influence of the obstacle position and the end position cannot be comprehensively considered, so that the obstacle avoidance can be realized, but larger deviation of the whole transportation path is easily caused, the driving distance or time is greatly increased, and the influence of the whole transportation process is larger. Disclosure of Invention The embodiment of the invention aims to provide a multi-view visual obstacle avoidance method and a system applying an image sensor, which aim to solve the technical problems in the prior art mentioned in the background art. The embodiment of the invention is realized as follows: The method for avoiding the obstacle by using the multi-view vision of the image sensor comprises the following steps: in the moving process, performing multi-view monitoring shooting of the environment, acquiring multi-view monitoring data, identifying the multi-view monitoring data, determining a target obstacle, and recording relevant position data; Performing obstacle avoidance analysis on the related position data, planning an obstacle avoidance selection area, and selecting a plurality of obstacle avoidance positions from the obstacle avoidance selection area; calculating the expected obstacle avoidance values of a plurality of obstacle-avoidance positions based on the related position data, comparing the expected obstacle-avoidance values with a preset threshold value, and judging whether an obstacle-avoidance condition is met; When the obstacle avoidance conditions are met, determining a final obstacle avoidance position from a plurality of obstacle-avoidance positions, and performing obstacle avoidance movement control; And when the obstacle avoidance conditions are not met, expanding and planning the obstacle avoidance selection area, determining the expansion selection area, and reselecting a plurality of obstacle avoidance positions. As a further limitation of the technical solution of the embodiment of the present invention, in the moving process, performing multi-view monitoring and shooting of the environment, obtaining multi-view monitoring data, identifying the multi-view monitoring data, determining a target obstacle, and recording relevant position data, the method specifically includes the following steps: In the moving process, carrying out multi-view monitoring shooting of the environment to obtain multi-view monitoring data; importing obstacle characteristic data; identifying the multi-view monitoring data based on the obstacle characteristic data, and determining a target obstacle; And carrying out positioning analysis on the multi-view monitoring data, and recording relevant position data. As a further limitation of the technical solution of the embodiment of the present invention, the positioning analysis is performed on the multi-view monitoring data, and recording the relevant posit