Grasping operation method of mobilemanipulator based on monocular vision
                                                                XIE Yushen, WU Qingcong, CHEN Bai, JIANG Meng
                  (College of Mechanical & Electrical Engineering, Nanjing University of Aeronautics & Astronautics, Nanjing 210016, China)

Abstract: Aiming at the problem of slow object recognition and low working accuracy of mobile manipulators, the object recognition and positioning algorithm and the robot operation accuracy improvement method were studied. Based on hardware such as omnidirectional mobile platform, industrial robot and monocular camera, a mobile manipulators system was constructed. The monocular visual template matching methods were summarized, and the feature point matching algorithm based on random tree classification was used to quickly and accurately identify and locate the objects. Camera calibration and robot handeye calibration was completed. The relationship between handeye pose and grab pose was analyzed. A compensation method for the correction of the handeye pose was proposed to reduce the working error. Finally, the grab/place experiment of the mobile manipulator was performed. The research results indicate that using the methods proposed above the objects are quickly and accurately identified and the working error is effectively reduced so that high working accuracy is achieved.

Key words: mobile manipulator; monocular vision; template matching; handeye calibration; error compensation