Robust Sliding Mode-Based Learning Control for Lane-Keeping Systems in Autonomous Vehicles

Abstract

In this paper, a robust sliding mode-based learning control (SMLC) scheme for lane-keeping systems (LKS) of road vehicles is proposed. It is assumed that all of signals in system satisfy Lipschitz-like condition, a robust sliding mode-based learning controller is designed to achieve the zero-error convergence of lateral position error dynamics. A new finding is that yaw angle error dynamics is able to converge to zero asymptotically on the sliding surface. Unlike many existing sliding mode control schemes, the proposed SMLC scheme does not require the bound information of unknown system parameters. More significantly, the LKS equipped with the SMLC algorithm exhibits a strong robustness against varying road conditions and external disturbances. Simulation results demonstrate that the designed SMLC scheme could exert excellent tracking performance and robustness.