Robust Tube-Based Model Predictive Control for Docking Process Self-Balancing Control of a Reconfigurable Unmanned Ground Vehicle

This study addresses the self-balancing control problem for a Reconfigurable Unmanned Ground Vehicle (RUGV) during high-precision docking process in complex environments. Constrained by underactuated dynamics and a narrow region of attraction, the self-balancing precision of RUGVs during docking readily degrades under model uncertainties, external disturbances, ultimately resulting in docking failure. To tackle this challenge, a novel Robust Tube-Based Model Predictive Control (RTMPC) framework is proposed for the self-balancing control of RUGV during docking. The RTMPC strategy provides an effective approach to explicitly deal with bounded disturbances and plant-model mismatches while rigorously enforcing both docking-error and actuator-input constraints. The core concept involves embedding a compact, precision-docking-tailored disturbance invariant set within a tube-based MPC scheme, enabling robust control of the desired docking attitude with negligible additional computational cost. In addition, the construction of the approximately minimal robust invariant set explicitly accounts for the complexity of the 4-dimensional coupled self- balancing model and the system’s underactuated nature. Due to the offline invariant set, compared to traditional MPC, the presented RTMPC achieves similar computation cost. The resulting RTMPC controller confines state trajectories in a tube centered around a nominal trajectory, effectively compensating for model uncertainties and external disturbances through error feedback, contributing significantly to the stability and robustness of the docking self-balancing process. Simulation results demonstrate that, compared to LQR and conventional MPC, the proposed RTMPC controller preserves vehicle stability and achieves high-precision, robust docking even under pronounced disturbances, thereby validating the effectiveness and superiority of the method.