Hierarchical Nonlinear Model Predictive Control for an Autonomous Racecar

This paper presents an optimization-based hierarchical motion planning and control architecture for autonomous racing that enables fast feedback times without significantly compromising on model accuracy. The online control scheme is divided into a high level NMPC for time-optimal trajectory planning and a low level NMPC for trajectory tracking using the same vehicle model. We numerically validate our approach in terms of computational efficiency and closed-loop performance in a detailed vehicle dynamics simulation environment. Despite considerable model-plant mismatches, the proposed hierarchical NMPC controller achieves a driven lap time in simulation which is only marginally slower than the theoretically optimal trajectory calculated offline via Optimal Control.