Model-free sliding mode control for path following in autonomous vehicle

This article addresses the challenges of parameter uncertainty and disturbance in the path-following task of autonomous vehicles. Instead of directly converging the path-following errors, a novel desired yaw angle function is introduced to enhance path tracking in underactuated vehicles, thereby simplifying the controller design. It is proven that the path tracking errors diminish to zero as the yaw angle aligns with the desired yaw angle. Based on the system model and control objectives, a composite model-free sliding mode control scheme is proposed. This strategy employs a non-singular terminal sliding mode control law to stabilize the control error. Furthermore, a high-order fast terminal sliding mode observer is incorporated to address system parameter uncertainty and external disturbance, with the estimated values utilized in the proposed controller. With this combination, this approach ensures tracking precision without requiring vehicle parameter knowledge and offers robust application, ultimately realizing model-free control. Meanwhile, the stability of the closed system is proven using Lyapunov theory. Finally, various operating conditions are designed to verify the robustness, and different control methods are compared to highlight the superiority of the proposed control strategy.