Adaptive Neural-Network-Based Fault Tolerant Control for Autonomous Vehicles With Dynamic Output Constraints

Vehicle reliability and safety are critical to the development and application of autonomous driving technologies. This article presents an adaptive neural network (NN) fault-tolerant controller design for autonomous vehicles, where dynamic output constraints are specifically considered to keep the vehicle within safety boundaries. First, considering the actuator faults, a vehicle dynamics model that incorporates system uncertainties and external disturbances is established. The fault-tolerant controller is then designed by integrating a radial-basis-function NN and a nonlinear disturbance observer to simultaneously address faults, uncertainties, and disturbances. Second, a barrier function is constructed to handle the vehicle output constraints, and a novel distance-based trigger function is introduced to activate the barrier function, enabling automatic smooth switching between constrained and unconstrained areas during path tracking. The direct Lyapunov method is subsequently employed to verify the boundedness of the closed-loop vehicle system as well as the parameter updating law. Finally, the effectiveness of the proposed control method is validated through both cosimulation and experimental tests.