Model Predictive Path-Following Control for UGVs with Curvature Inaccuracy: A Disturbance Rejection Approach Combining Speed Regulation

In this paper, a disturbance rejection model predictive control (MPC) framework is developed for the path-following task of unmanned ground vehicles (UGVs) with curvature inaccuracy. First, a modified path-following model is established to show the positive correlation between vehicle speed and the influence of curvature inaccuracy. Second, with the inaccuracy regarded as the external disturbance, an extended state observer is designed for estimation and the observed result is compensated into the control channel. Third, a robust MPC approach is developed to get the optimal angular velocity under several safety and comfort constraints. Besides, by incorporating a penalty term for speed to the loss function, this MPC algorithm also possesses the capability to adjust speed according to curvature inaccuracy. Finally, the feasibility and stability of the disturbance rejection MPC framework are guaranteed. The efficacy and advantage of the proposed approach are confirmed through comparative experiments under different curvatures.