基于RoboMaster竞技赛中机器人视觉识别定位系统的研究

Research on Robot Visual Recognition and Positioning System in RoboMaster Competition

针对RoboMaster竞技赛规则,设计一款具有自主识别与定位视觉系统的移动机器人,用于识别敌方机器人特定目标,从而实现精确定位与打击。系统采用基于C++语言的OpenCV相关算法,运算平台搭建多线程任务并发执行框架来提高系统运算效率,对识别目标的视觉元素、颜色特征及目标形态进行循环二值化阈值滤波算法,获得目标的精确特征参数;使用PnP目标解算方法对特征信息定位的准确性和实时性进行优化,定位数据经串口通信发送给STM32控制系统处理。这些措施有效地提高了比赛中对敌方机器人的特定目标识别定位的精确性与实时性。实验结果表明,采用综合二值化滤波算法的定位技术有效提高了匹配精度,满足了移动机器人快速识别与定位的需求。

A recognition and positioning system was designed with OpenCV related algorithm based on C + + in RoboMaster competitionto accurately locate and attack specific targets. A high efficiency computing platform was built with multi-threaded task concurrent execution framework in the system. The visual elements,color features and target shape of the recognized target were analyzed,and the circular binary threshold filtering algorithm was used to obtain the accurate characteristic parameters of the target. The system optimizesthe accuracy and real-time of feature information positioning through PNP target solution method,and the positioning data aresent to STM32 control system for processing through serial communication,which effectively improvesthe accuracy and real-time of competitive robot’s recognition and positioning of enemy robot’s specific target in the competition. The experimental results show that positioning technique and circular binary threshold filtering algorithm can effectively improve the matching accuracy,which meets the requirement of fast recognition and localization of mobile robot.