A Simulation Research on 3D Visual Servoing Robot Tracking and Grasping a Moving Object

Abstract

Visual servoing provides a great convenience for flexible robotic systems. However, there is contradiction between enlarging the working field and improving the accuracy of manipulation. In this paper, a new position based visual servoing approach is presented to the dynamics manipulation of moving objects. The research is divided into three aspects: At first, a global-local vision tracking method is proposed. The vision system includes two cameras, one fixed in the workspace as a global visual system, the other mounted at the end-effector as a local visual system. In every sample interval, the 3D positions of feature points of objects are obtained by the two-camera stereo vision system. And a dynamic extended Kalman filter (EKF) is employed to estimate the kinetic parameters. Secondly, a biomimic motion planning algorithm of manipulator is developed. The configuration of the manipulator in our research is a 5DOF manipulator alike to the arm of human being including a 2DOF shoulder, a 1DOF elbow and a 2DOF wrist. According to the configuration, a motion planning algorithm is proposed which consists of the planning of the wrist and the orientation of the gripper. Finally, a two stages task planning including tracking stage and grasping stage is developed. Simulation results show the feasibility of this approach.