Robust Predictive Steering Control for Autonomous Vehicles with Polynomial Noise Resilience Neural Dynamics

Autonomous driving systems are inevitably influenced by kinds of noise, which may further incur unwanted safety problems. In this paper, a polynomial noise resilience neural dynamics (PNRND) is proposed to efficiently solve the predictive steering control (PSC) strategy for autonomous vehicles. In this sense, the reliability of autonomous vehicles can be guaranteed under noisy conditions, which are potentially caused by uneven road surfaces and sensor inaccuracies. By applying the Taylor expansion theorem, these noises can be represented in polynomial forms for the desired accuracy, providing a unique perspective for handling noise disturbances. Moreover, the robustness and stability of the proposed PNRND model are analyzed using theoretical derivations. At last, experiments conducted on the CarSim-Simulink platform and hardware-in-the-loop (HIL) platform demonstrate the effectiveness, robustness, and real-time performance of the proposed neural dynamics approach.