Precise Positioning of a Robotic Arm Manipulator Using Stereo Computer Vision and Iterative Learning Control

Modern service robots are a combination of a mobile platform and a robotic manipulator. One of the main and most difficult tasks in front of these robots is the object transportation between two points. They should be able to detect any desired object. Then move the platform as close as possible to the desired object. Afterwards the manipulator arm should position the gripper near it. The last step is to grasp and transport the object. This paper presents a novel approach for solving the manipulator arm positioning problem. The presented method combines computer vision and Iterative Learning Control techniques in order to compensate any imprecisions of the robot kinematics and dynamics. This results in an efficient solution, which succeeds in precise positioning near the desired object even when there is a very little knowledge of those mathematics models. It is a robust method, which auto adapts to mechanical wear during normal operations, not severe damages or imprecise factory assembly. The method is then validated on a physical robotic manipulator.