Active Steering Controller for Driven Independently Rotating Wheelset Vehicles Based on Deep Reinforcement Learning

This paper proposes an active steering controller for Driven Independently Rotating Wheelset (DIRW) vehicles based on deep reinforcement learning (DRL). For the two-axle railway equipped with Wheelsets (IRWs), each wheel connected to a wheel-side motor, Ape-X DDPG controller, enhanced version of Deep Deterministic Policy Gradient (DDPG) algorithm, is adopted. Incorporating Distributed Prioritized Experience Replay (DPER), trains neural network function approximators obtain data-driven DIRW controller. utilized control input torque wheel, aiming improve capability IRWs. Simulation results indicate that compared existing model-based H∞ algorithm and demonstrates better curving performance centering ability in straight tracks across different running conditions significantly reduces wheel–rail wear. To validate proposed algorithm’s efficacy real vehicles, 1:5 scale model vehicle its digital twin dynamic were designed manufactured. The was deployed subjected experiments scaled track. experimental reveal under both curved enhanced.