Extending Visual Servoing Techniques to Nonholonomic Mobile Robots

The stabilization to a desired pose of a nonholonomic mobile robot, based on visual data from a hand-eye system mounted on it, is considered. Instances of this problem occur in practice during docking or parallel parking maneuvers of such vehicles. In this paper, we investigate the use of visual servoing techniques for their control. After brie y presenting the relevant visual servoing framework, we point out some problems encountered when it is considered for nonholonomic mobile robots. In particular, simple velocity control schemes using visual data as feedback cannot be applied anymore. We show how, by using the extra degrees of freedom provided by the hand-eye system, we can design controllers capable of accomplishing the desired task. A rst approach, allows to perform a visual servoing task de ned in the camera frame without explicitly controlling the pose of the nonholonomic mobile basis. A second approach based on continuous time varying state feedback techniques allows to stabilize both the pose of the nonholonomic vehicle and that of the camera. The experimental evaluation of the proposed techniques uses a mobile manipulator prototype developed in our laboratory and dedicated multiprocessor real time image processing and control systems. To appear in: The Con uence of Vision and Control, Eds. G. Hager, D. Kriegman, S. Morse, Lecture Notes in Control and Information Systems, Springer-Verlag, 1998.