Dual-Mode Model Predictive Motion Cueing

Dynamic driving simulators rely on motion cueing algorithms to generate platform trajectories that reproduce accelerations and rotations of simulated vehicles within physical limits. The employment of model predictive control (MPC) has emerged as a promising approach for this task, primarily due to its capability to address constraints explicitly. However, ensuring closed-loop stability poses a fundamental challenge. Conventional stability-enforcing mechanisms, such as terminal equality constraints or terminal costs, limit workspace utilization by inducing premature washout behavior, while extending the prediction horizon conflicts with stringent real-time requirements. This paper presents a dual-mode architecture that decouples stability certification from performance optimization. A safety mode based on linear state feedback defines a maximal positive invariant set that guarantees asymptotic stability and constraint satisfaction. Within this set, a performance mode employs MPC for reference tracking without terminal constraints, maximizing workspace utilization. The approach is formulated as a quadratic program and validated on a robot-based driving simulator. Simulation results confirm guaranteed constraint satisfaction and safe mode switching capability. The architecture reconciles stability certification with performance optimization for MPC-based motion cueing.