Dynamic visual servoing of robotic manipulators

This paper proposes a new controller for Dynamic Visual Servoing of robotic manipulators. In this work the image is accquired from a CCD camera mounted on a planar robotic manipulator end-effector, and processed in a standard PC based vision system in order to close the Visual Servoing loop with PD controllers. First, we propose an extension to Dynamic Visual Servoing based on a pose estimation algorithm and then compare it with another Lyapunov stable controller. The proposed dynamic controller uses only information from the image and robot encoders. This approach is inspired on classical 3D Visual Servoing and the Jacobian transpose controller approach. An experimental setup developed for visual servoing experiments, including software and hardware is presented. Experimental results for a planar robotic manipulator are presented.