Data-Driven Control for Vehicle Platoon System with Actuator Faults and External Disturbances

This paper presents a decentralized control framework for vehicle platoon systems designed to address the challenges posed by actuator faults and external disturbances in real-world scenarios. Existing approaches, such as linear feedback and model predictive control, are often limited by their sensitivity to dynamic faults, dependence on precise system parameters, and fragmented handling of faults and disturbances. These limitations can undermine both safety and operational efficiency. To overcome these issues, we propose a unified control architecture that seamlessly integrates fault and disturbance compensation, eliminating the need for explicit fault detection or disturbance estimation. By reformulating the problem as a multi-objective cost minimization task, adaptive control laws are derived using online reinforcement learning, thereby ensuring robustness against parameter uncertainties and time-varying anomalies. We provide theoretical guarantees for closed-loop stability and string stability, specifically for vehicle platoon systems.