Image-based visual control of nonholonomic mobile robots via velocity fields: Case of partially calibrated inclined camera

This paper addresses the wheeled mobile robot guidance under fixed-camera configuration by means of image-based velocity fields. Two practical ingredients are considered in this paper: first, the image lane may not be parallel to the plane where the robot moves, and second, it is assumed only partial camera calibration; more specifically, with regard to the camera position, only its height is assumed to be known. A control system - with embedded desired velocity field - is proposed ensuring that the robot follows asymptotically the flow lines. This is proven by invoking the LaSalle's invariance principle. The effectiveness of this approach is demonstrated by experimental results