Robust Lane Keeping Control with Estimation of Cornering Stiffness and Model Uncertainty

Abstract

This paper introduces an adaptive lane-keeping control strategy that adapts to varying cornering stiffness while ensuring robustness against uncertainties. The system consists of three blocks: an Interacting Multiple Model (IMM) cornering stiffness estimator, a cornering stiffness uncertainty estimator, and a Robust Model Predictive Controller (RMPC). Improvements in estimation accuracy are achieved through a novel IMM probability derivation method, and the uncertainty estimator utilizes the IMM probability matrix to obtain reliable uncertainty boundaries. Real-time cornering stiffness estimations are integrated into the RMPC for adaptive model predictions. Uncertainty boundaries provide robustness against estimation error in the RMPC by constraint tightening and smoothing techniques. The performance of the estimator and controller is validated in simulations, where the overall control performance is compared to that of the Model Predictive Control (MPC) based on static cornering stiffness.