Dynamically Grasping with Incomplete Information Workpiece Based on Machine Vision

In automatic feeding process, there exists the problem of incomplete image information of workpiece, which brings difficulties to pose detection and dynamic grasping. This paper proposed an algorithm of Characteristic Region Minimum Rectangle Fitting to achieve the pose of assembly workpiece with incomplete information, which was based on the camera-conveyer-robot hybrid modular calibration. Then combined pose estimation on the conveyor to realize dynamic and stable grasp. The effectiveness and accuracy of the proposed method are further demonstrated by experiments of mobile phone assembly using LINKHOU robot. The results showed that the proposed algorithm can realize workpiece positioning in automatic feeding process, and the positioning and recognition accuracy of the workpiece can complete the dynamically grabbing task of the robot arm.