Robot visual servoing by direct image feedback

This paper analyzes the robot hand-eye coordination from the view point of integration of vision and control. A new scheme of direct image feedback control is proposed. It compares the real time sampled image with the desired one and forms an error image. With this 2-D error image, The control action steers the gripper to move until the image error is eliminated. Considering the local minimum problems that arise with multimodal image error, we adopt fast simulated annealing for global searching.