Steering overshoot suppression in autonomous vehicles via a magnetorheological assisted steering system

Abstract Steering performance is critical for accurate and safe path tracking in autonomous vehicles (AVs), yet actuator overshoot during high-rate maneuvers can induce significant tracking errors.Most existing studies focus on improving high-level path-tracking algorithms and assume ideal steering actuator behavior, which may not hold at elevated steering rates. To address this limitation, this study proposes a magnetorheological-assisted steering (MRAS) system that integrates a compact rotary magnetorheological damper (MRD) into an automated electric power steering (EPS) system to provide current-controlled damping torque at the steering motor. A semiactive control algorithm based on an extreme learning machine (ELM) schedules MRD current as a function of steering rate to suppress overshoot while preserving command-tracking responsiveness. A numerical model of the MRAS is developed, and the proposed approach is evaluated using step and sweep input profiles in both numerical simulations and hardware experiments. The results show that MRAS reduces steering overshoot compared with a baseline EPS system, while preserving rise time and settling time. These findings demonstrate that semi-active, MRD-based damping can enhance low-level steering control performance and effectively complement high-level path-tracking controllers in autonomous vehicles.