Autonomous Vehicle Path Tracking: Differential Lyapunov Optimal Control and Decision Fault Tree Learning

Precise motion control is critical for guiding self-driving cars or autonomous vehicles, posing significant challenges in tasks such as nonlinear trajectory tracking and lane changing. A reliable and highly efficient steering mechanism is essential, even under challenging conditions. This paper proposes an optimal control strategy using a differential Lyapunov approach with adjustable Z-matrices to design a path-tracking system that follows a reference trajectory. The method effectively addresses parametric uncertainties related to vehicle cornering stiffness, road tire adhesion, inertial parameters, forward speed, as well as external disturbances, measurement noise, and localization or path-planning faults. To account for autonomous vehicle constraints, an alternative trajectory generation model is developed to ensure safety. A fault detection technique called Decision Fault Tree Learning (DFTL) is introduced to identify and locate faults within large and complex communication networks. Extensive testing demonstrates the method's high accuracy. Simulations and comparative analyses with advanced methods validate the fault detection rate and response time, emphasizing the technique's effectiveness.