Model Predictive Control for Autonomous Vehicle Tracking

Abstract

This study develops model predictive control (MPC) schemes for controlling autonomous vehicles tracking on feasible trajectories generated from flatness or polynomial equations. All of the vehicle online moving parameters including coordinate positions, body orientation angle, and steering angle are included into the MPC optimizer for calculating the real-time optimal inputs for the vehicle linear velocity and its steering velocity to minimize the errors between the desired and the actual course of travel. The use of MPC can simplify and eliminate the complexity of controller design since MPC can work itself as a system modelling controller. MPC can also handle online the constraints of any variables exceeding their limits. However the high computational demands are the main challenge for this method applying for the real applications.