Path following of an unmanned ground vehicle with GPS feedback using model predictive control method

In this paper, path following control of an unmanned ground vehicle is presented based on the feedback of position and orientation errors. The measurement unit of this autonomous vehicle prototyped for such tasks contains a real-time kinematic global positioning system (RTK-GPS), an inertial measurement unit (IMU) and an absolute encoder to accurately define the position and orientation of the car. A model predictive control was proposed for the path following of the mobile robot based on the successive linearized and discretized kinematic model. This optimal control method performs on the lowest position and orientation errors with respect to a non-holonomic virtual vehicle that is considered to move flawlessly on a given reference path and the smoother steering angle. The paths followed here are defined by rationally based splines or known geometric curves created with control points from a digital mapping program. This paper includes both simulation and real-time experimental studies. The outcomes were examined in terms of the design performance and control strategy of the vehicle. Despite the physical constraints on the vehicle prototype, it has been observed that position and orientation errors occur within satisfactory limits. In particular, the fact that the steering angle is not subjected to excessive oscillations indicates that the control method has a good performance.