关键词: 机器人位姿测量, 视觉测量, 特征识别, 主动光立体标靶

Abstract: A visual method with an active light target was proposed to obtain the positions and orientations of robots herein. First, the active light stereo target adopted active infrared light instead of reflective markers and employed shape features instead of brightness features to encode the location informations. Second, the stereo structure was designed to provide the spatial geometric constraint information. The active light target owned five feature planes to expand the visibility no mater what pose. Third, a detection method of feature planes was proposed. The proposed feature detection method was testified under different situations, such as indoor illumination environments, the direct solar radiation at outdoor square and the mixed illumination environments. The experimental results confirmed the feasibility and reliability of the proposed method. Forth, the precision of the proposed visual method was verified in a high precision platform. The experimental results show the average position error is as 0.0857mm and the average orientation error is as 0.1086° in measuring distance ranging from 0.5~3m. Finally, the pose measurement of a YASKAWA robot arm, Motoman-MH80, was conducted and the experimental results confirmed the feasibility and efficiency of the proposed method.

Key words: full pose measurement of industrial robot, visual measurement, feature detection, active light stereo target