Real-time driving mode advice for eco-driving using MPC

This paper proposes an on-line advice eco-driving assistance system (EDAS) for providing the optimal velocity profile to improve fuel economy. The EDAS employs a driver-in-the-loop (DIL) framework, where an adviser is designed to provide high-level driving mode suggestions while the low-level control commands such as throttle and brake, are left to the driver to implement. A simplified dynamic model is developed in the adviser excluding continuous-time control variables such as the engine torque and engine brake torque. The adviser employs an event-triggered model predictive control (MPC) algorithm to provide suggestions in real-time using predictive road and traffic information. On-line computational cost for the MPC has been significantly reduced using an efficient mixed-integer optimal control (MIOC) algorithm. To demonstrate the efficiency and effectiveness of the proposed EDAS, a numerical study and a simulation using measured data from a real-life driving test is conducted. Comparisons are made between the proposed EDAS and an eco-driving controller considering both high and low level control inputs without a driver.