Synthesis of Robust Model Predictive Control for Vehicle Path Following with Uncertainties

In the problem of path following of autonomous vehicles, the complexity of vehicle model and road surface will bring time-varying parametric and unstructured uncertainties to the system. In our study, we design a controller based on synthesis of model predictive control method, which is robust to the uncertainties of vehicle longitudinal velocity, tire cornering stiffness and path curvature. According to the vehicle dynamics model, the vehicle linear uncertain system model is established, where road curvature is modeled as system state rather than system noise. A vehicle path following controller based on synthesis robust model predictive control (SRMPC) method is proposed, which is robust to uncertain parameters and has low online computational expense. The results of co-simulation conducted on Prescan and MATLAB/Simulink show that, compared with offline robust model predictive control method, the lateral error of path following controller based on SRMPC method is smaller, and the robustness is better.