摘要
能被机器人识别的特殊标记被广泛应用于移动机器人的工作空间内,通过实时识别与计算可获取移动机器人的位姿。但是,人眼无法直接从以条形码、二维码为代表的特殊标记中获得信息,在人机配合场合,往往需要多套标记系统实现指示,导致系统趋于复杂。为了解决移动机器人在室内环境下的自定位问题,文中提出一种基于人机共识标记的室内移动机器人单目视觉定位的方法,采用单目摄像头作为信息采集单元,实时检测并识别预先布置的人机共识标记,获取标记角点的世界坐标信息;然后利用基于单帧图像特征点匹配的P4P算法实现机器人的定位。通过理论分析得出,所提方法能够快速完成机器人的位姿计算,相较于传统的人工标记定位,新型人机共识标记具有结构简单、交互性强、实施灵活等优点。
Special markers that can be recognized by robots are widely used in the workspace of mobile robots,and the pose of mobile robots can be obtained by means of real-time recognition and calculation. However,the human eye cannot directly obtain information from special marks represented by bar codes and two-dimensional codes. In the case of human-machine cooperation,it often needs multiple marking systems to realize instructions,resulting in the system becoming more complex. In order to solve the self-positioning problem of mobile robot in indoor environment,a monocular vision positioning method of indoor mobile robot based on man-machine consensus marks is proposed. The monocular camera is used as the information acquisition unit to detect and identify the pre-arranged man-machine consensus marks in real time,and obtain the world coordinate information of the marked corners. The P4P algorithm based on single-frame image feature point matching is used to realize robot positioning. The heoretical analysis results show that this method can quickly complete the pose calculation of the robot. In comparison with the traditional manual marker positioning,the new man-machine consensus marker proposed in this paper has the advantages of simple structure,strong interaction,and flexible implementation.
作者
郭敏
王超
范青荣
吴超群
刘辉
陈瑜
GUO Min;WANG Chao;FAN Qingrong;WU Chaoqun;LIU Hui;CHEN Yu(Institute of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan 430070,China;Wuhan Deep-sea Yizhi Technology Co.,Ltd.,Wuhan 430024,China;Center of Productivity Promotion in Beijing,Beijing 100089,China)
出处
《现代电子技术》
2022年第20期183-186,共4页
Modern Electronics Technique
关键词
人机共识标记
移动机器人
单目视觉
室内定位
图像处理
图像识别
特征点匹配
human-machine consensus marker
mobile robot
monocular vision
indoor positioning
image processing
image recognition
feature point matching
