Adaptive vision-based path following control of a wheeledrobot

This paper addresses the problem of vision-based path following control for a wheeled robot of the unicycle type. A nonlinear controller is derived that yields convergence of the vehicle´s position to a straight line in the presence of model parameter uncertainty. Control design starts with a kinematic model of the system that describes the relationship between robot inputs and the pose of the robot with respect to the path directly in the camera plane, using selected path features. A kinematic control is then derived to drive path following errors to zero. The kinematic controller is further extended to cope with vehicle dynamics by resorting to backstepping and Lyapunov based techniques. Robustness to system parameter uncertainty is addressed by incorporating a parameter adaptation scheme. Simulations results illustrate the performance of the controller derived.