Abstract
Outdoor mobile robots must perform operations with ever-increasing speed and distance. Therefore we are interested in designing controllers of fast rovers which improve mobile robot capacity in natural environment. When designing autonomous path tracking systems for fast rovers, a major problem is the dynamic effect and the non-linearity of the model. Several control laws have been designed to resolve the problem by separating the dynamic of the robot at the problem of trajectory tracking.
This paper presents a path tracking controller for a fast rover with independent front and rear steering. In the first step, a dynamic model of a vehicle that moves on a horizontal plane was developed. Next, the projection of the position of the vehicle in the absolute reference frame was used to define the kinematics non-linear model. We present a new approach to solving a tracking path problem by applying Non-linear Continuous-time Generalized Predictive Control (NCGPC). The controller is based on the dynamic model of a bicycle like vehicle which considers the lateral slippage of the wheels. The prediction model allows anticipation of future changes in setpoints in accordance with the dynamic constraints of the system. Experimental results, show a good control accuracy and appears to be robust with respect to environmental and robot state changes.