Research On Robot Positioning And Moving Target Detection Algorithm In Competition System

As a participatory AI challenge project,robotics competition is achieved through specific mission objectives,enabling researchers to use a variety of techniques and ingenious ideas to achieve better solutions to specific problems.It can promote the exchange of scientific research units,universities and enterprises,and can promote the progress of academic research.Under the background of the implementation of China's artificial intelligence strategy,researching robot competition is a topic with important practical application significance.The robot competition system is a system that relies on real-time comparison,and has high requirements for positioning and navigation of the robot.Not only does it require precise positioning,but it also needs to be able to detect moving robots.In this paper,aiming at how to realize the rapid positioning and moving target detection of multiple robots in the competition system,combined with the research data at home and abroad,the visual background extraction algorithm Vi Be(Visual Background Extractor)combined with ORB(Oriented Fast and Rotated Brief)feature matching is proposed and improved.The defect of the original algorithm completes the rapid positioning of the robot in the competition system.Mainly done the following research work:(1)Simplified the monocular positioning model according to the requirements and characteristics of the robot competition system,and designed a monocular vision positioning scheme to meet the application of the competition system scene.Finally,the feasibility of the positioning scheme was verified by camera calibration experiments.(2)In this paper,a ghost elimination algorithm for the positioning problem of moving robots in the competition system is designed.This algorithm is based on the static background characteristics of the competition system.The moving foreground image extracted by Vi Be is matched with the sample image by ORB feature.Whether there is ghosting in this area.(3)In this paper,the ORB feature matching algorithm is used to realize the target recognition of robots in the competition venue.In order to improve the number of feature points in the edge region and the texture-poor region in the competition scene,the problem is insufficient and the extraction is difficult.The quadtree block method is used to image.Perform block processing.(4)In the competition system,the motion robot often crosses the quadtree static block boundary line,which leads to the problem of ORB feature point extraction and matching error.This paper uses Vi Be detection result to dynamically adjust the quadtree block boundary line position and reduce the robot.The influence of the matching result on the edge of the dividing line or across the boundary line improves the matching accuracy of the ORB feature points of the moving robot in the competition system.In this paper,the improved Vi Be algorithm is used to verify that the improved algorithm can effectively reduce the influence of ghosts on the detection results,and can adapt to the phenomenon of multiple light mutations in the competition venue.The moving target recognition experiment of the competition robot proves that the proposed method can meet the recognition requirements of the robot in the competition system.