MPC-based lane keeping for commercial vehicles with steering dead zone consideration

Abstract

Commercial vehicles face challenges such as a large steering dead-zone and difficulties in directly obtaining lateral velocity through sensors. To address these issues, a model predictive control (MPC) lane keeping algorithm is designed, taking into account the characteristics of the steering dead-zone. First, a vehicle dynamics model and state error equations are established. Then, the vehicle status is estimated using an extended Kalman filter (EKF). Based on the estimated vehicle parameters, an MPC lane keeping controller is designed. Finally, depending on the different operating states of the vehicle, the compensation strategies are designed based on the yaw rate error and the dead-zone length. The lane keeping algorithm was validated through simulation experiments. The lateral velocity estimation error was less than 4.8%, and the influence of the steering dead-zone decreased by 8 degrees. The results indicate that the improved MPC lane keeping algorithm exhibits excellent tracking accuracy. The dangerous S-shaped trajectory of the vehicle is effectively avoided, and the stability of vehicle operation is improved.