Computationally Efficient Observer-Based Robust MPC for High-Speed Vehicle Centralized Trajectory Tracking With H ∞ Disturbance Attenuation

This paper proposes a novel centralized trajectory tracking LPV controller for automated vehicles based on Observer-Based Robust Model Predictive Control (OB-RMPC), combining longitudinal and lateral control of the vehicle through a coupled dynamics vehicle model with the aim of applying it during high-velocity evasive maneuvers. For this purpose, minimization of the trajectory tracking errors has been pursued through a cost function which combines the MPC estimated states and control input costs with anH∞condition. Since not all vehicle states are available for easy measurement, an observer-based approach has been followed for the controller synthesis. First, a Luenberger observer is designed for unknown states estimation. Then, the OB-RMPC optimization problem is tackled by addressing the dynamics of the estimation error and estimated states. Quadratic boundedness condition, augmented system robust positive invariance, and actuator saturation constraints are managed through the use of Linear Matrix Inequalities (LMIs) formulation. A mechanism to update the stability boundaries of the estimation error set is implemented online. Finally, to mitigate the computational demands of the OB-RMPC strategy, an online algorithm is proposed. The control action is calculated based on the states estimation, the optimized bound of the estimation error and an offline-constructed look-up table. The proposed strategy is validated through simulations in Matlab®Simulink and CarSim®under different high-velocity driving scenarios. The obtained results emphasize the superior tracking effectiveness of the proposed control method compared to other state-of-the-art control strategies in high-velocity maneuvers without compromising lateral stability.