A Two-Time-Scale Control Scheme for Long-Horizon Vehicle Motion Control

There is a conflict between the long-term control horizon and the short-term control step due to the computational burden. In addition, the significant burden associated of controller tuning and noise rejection must be considered. According to different time scales between the translational and rotational motions of the vehicle, this paper proposes a two-time-scale control scheme to handle the conundrum. The computing load is lessened through transforming the trajectory tracking problem into two successive finite-horizon optimal control problems. In the outer-loop controller, the translational motion is regulated over an extended control horizon to generate a human-like tracking trajectory. In the inner-loop controller, the rotational motion is regulated based on a high-fidelity model to guarantee precise manipulation. Additionally, taking the side-slip angle into consideration, an MPC based output regulator (ORMPC) is proposed to improve the steering response to sharp curves. Theoretical analyses indicate that the proposed method provides a transient profile for trajectory tracking and ORMPC guarantees the convergence of yaw angle in curves. As a result, lateral jerk is reduced significantly and steering smoothness is improved. Simulation and experimental results demonstrate that the proposed method reduces the lateral jerk by an average of 50% and improves the ride comfort, parameter adaptability and noise rejection in different driving scenarios, compared with conventional methods.