Adaptive model predictive control based on stability index for path tracking of autonomous vehicles considering model error

Abstract

Autonomous vehicle path tracking control based on model predictive control (MPC) still faces some challenges, such as unknown external environment disturbance and parametric modeling uncertainties. For these problems, this paper puts forward an adaptive model predictive control (AMPC) strategy which combines stability index and model error compensation to improve vehicle path tracking and stability performance. Firstly, the model error identification scheme based on radial basis function neural network (RBFNN) is designed to estimate the uncertainties of vehicle parametric model. Through the identification and compensation of model error, the model error during critical maneuvers can be reduced. Secondly, to improve vehicle maneuverability and yaw stabilization performance, the AMPC controller is realized by adaptive weight adjustment based on MPC. The stability index is designed to schedule the weights of MPC optimization objectives. Finally, the effectiveness and robustness of the path tracking control strategy are verified by CarSim&Simulink co-simulation and hardware-in-the-loop (HIL) test. The results illustrate that the AMPC strategy based on stability index considering model error not only enhances the tracking accuracy, but also improves the handling and stability performance of vehicles.