Trajectory Tracking of Autonomous Driving Vehicles via Output Feedback MPC

Abstract

This paper aims at addressing a model predictive control technique for trajectory tracking of autonomous driving vehicle under the condition that the system states are difficult to be measured. Firstly, a linearized error model and an dynamic output feedback controller are established, and then an augmented closed-loop model is obtained to predict the dynamic behavior of autonomous driving vehicles. In addition, the dynamic output feedback model predictive controller is derived by minimizing the upper bound of an infinite horizon quadratic objective function which explicitly takes the input constraint into account. Finally, the optimization problem of dynamic output feedback model predictive control (DOFMPC) is solved by linear matrix inequalities (LMIs) technique, and the first one of the optimized control sequence is implemented. It is proved that the output feedback predictive control strategy realizes the goal of accurately tracking the reference trajectory on the simulation platform of CarSim and Simulink, and the effectiveness of the algorithm is verified.