Finite-time path following tracking control for time-delayed autonomous vehicles using an enhanced RBF neural network

In the field of autonomous driving, accurate path-following control is crucial for ensuring safety and efficiency. This paper addresses the challenges of path-following errors and convergence speed in autonomous vehicles by proposing a novel SOSM-RBF-TD-FT control strategy. Firstly, a comprehensive system model was established, which considered disturbances and vehicle time delays. Secondly, based on Second Order Sliding Mode (SOSM) control theory, a Radial Basis Function Neural Network (RBFNN) was utilized to approximate system nonlinearity and improve control accuracy. Subsequently, a finite-time control scheme was introduced to further enhance following accuracy and improve the dynamic performance of the system. Finally, the proposed method was validated through MATLAB simulations and real vehicle platform experiments. Simulation results demonstrated that, compared to the SOSM method, the SOSM-RBF-TD-FT method significantly reduced the lateral offset by 1.8813 m and the 2-Norm error by 30.7043 m. On a real vehicle platform, the proposed method further reduced the maximum and mean lateral offsets by 46.92% and 75.79%, respectively. The results showed that the SOSM-RBF-TD-FT method significantly decreased the path-following error and achieves faster convergence, which provides a theoretical basis and methodology for the development of path-following control technology for autonomous driving.