Fuel-efficient predictive cruise control using the explicit MPC method for commercial vehicles

Fuel-efficient predictive cruise control (FPCC) is of great significance in achieving fuel conservation. Model predictive control (MPC) serves as a promising method for the design of the FPCC controller. However, existing MPC-based FPCC controller on real vehicles remains challenging since MPC needs to find the optimal control law at each time step with limited computation time and resource. In this paper, we propose a learning-based explicit MPC method to learn the optimal policy of FPCC systems. We employ the neural network to approximate the policy, and transfer the online computation burden of the optimal control law to the offline policy training process. Simulations demonstrate that the method can effectively improve the real-time performance and be generalised to different road topologies without sacrificing fuel economy and travel efficiency.