基于单目视觉定位与运动目标检测算法的机器人在竞赛系统中的研究  

Research on Robots in Competition Systems Based on Monocular Vision Localization and Moving Object Detection Algorithms

作　　者：计大威 邱洋[1] JI Dawei;QIU Yang(Shanghai Technical Institute of Electronics and Information,Shanghai 201411,China)

机构地区：[1]上海电子信息职业技术学院,上海201411

出　　处：《自动化与仪器仪表》2024年第4期201-205,共5页Automation & Instrumentation

基　　金：《交通强国战略背景下动车组检修专业课程思政探究与实践》(203760)。

摘　　要：机器人竞赛系统具有很强的实时性,该系统不但要有准确的定位能力,而且要有对移动机器人进行检测识别的能力。研究针对学生机器人竞赛中的定位与检测问题,在特征匹配的视觉背景提取算法(Visual Background Extractor,ViBe)的基础上,结合方向性FAST特征点检测和旋转BRIEF描述子,改进竞赛系统,期望实现对学生竞赛系统的准确定位。结果显示,改进后的ViBe算法准确率最高,平均准确率为87.34%;改进后的ViBe算法误检率最低,平均误检率为9.12%。改进后的ViBe算法能够有效识别鬼影并且进行消除处理,同时能消除光照突变带来的影响,有效定位与识别机器人。结果表明研究方法具有优越性,且能够提升匹配速度,对学生竞赛系统中机器人的定位与运动目标的检测具有重要意义。The robot competition system has strong real-time performance,which not only requires accurate positioning ability,but also the ability to detect and recognize mobile robots.Research focuses on the positioning and detection issues in student robot competitions.Based on the Visual Background Extractor(ViBe)algorithm for feature matching,combined with directional FAST feature point detection and rotating BRIEF descriptors,the competition system is improved to achieve accurate positioning of the student competition system.The results show that the improved ViBe algorithm has the highest accuracy,with an average accuracy of 87.34%;The improved ViBe algorithm has the lowest false detection rate,with an average false detection rate of 9.12%.The improved ViBe algorithm can effectively recognize ghosts and perform elimination processing,while also eliminating the impact of sudden changes in lighting,effectively locating and recognizing robots.The results indicate that the research method has superiority and can improve the matching speed,which is of great significance for robot localization and motion target detection in student competition systems.

关 键 词：机器人 竞赛系统 定位 ViBe算法 ORB特征匹配 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]