Human-Machine Shared Steering Control For Vehicle Lane Keeping System: An Observer-Based Model Prediction Control Method

The paper introduces a human-machine cooperative lane-keeping assistance driving system that relies on a Unknown Input Observer (UIO). In contrast to the traditional Linear Parameter-Varying (LPV) model, this paper simplifies the LPV polyhedral model using a polyhedral simplification approach. The UIO is designed with unknown input decoupling technology to ensure asymptotic estimation performance for vehicle dynamics and driver torque. Based on the estimation results of the unknown input observer, this study proposes a UIO-based model predictive controller (MPC) that will enable the vehicle to track the road centreline with its accuracy. Additionally, a human-machine cooperative control authority distributor is introduced to mitigate conflicts between human and machine control. The proposed method’s effectiveness is confirmed by co-simulation using Simulink-Carsim, demonstrating its performance.