A Two-Layer Approach for Predictive Optimal Cruise Control

Optimization-based strategy planning for predictive optimal cruise control has the potential for significant improvements in passenger comfort and fuel efficiency. It is, however, associated with a high computational complexity that complicates its implementation in an electronic control unit. When implementing predictive cruise control, real-time capability must be ensured while maintaining optimal control performance in the presence of disturbance and model uncertainty. Real-time capability can be achieved either by a significant simplification of the optimization problem or by a layered control approach, combining the strategy planner with a low-level controller. Both approaches, however, are prone to deteriorate optimal control performance, particularly in the presence of disturbance. We present a model-predictive controller structure that extends the layered control approach by using the same optimization algorithm on two layers. A low-frequency planner that generates the optimal control strategy is combined with a high-frequency stabilization planner that tracks the strategy and closes the control loop on a short planning horizon. This reduces the computational load while maintaining an optimal response to disturbance. The approach is applied to a predictive cruise control system and compared to existing stabilization schemes in a simulation environment.