Learning-Based Collaborative Speed Planning and Energy Management of Electric Vehicles Among Signalized Intersections

Energy management and eco-driving are crucial for the energy saving of range-extender electric vehicles (REEVs) in urban traffic environments. In this work, the collaborative energy management and vehicle speed optimization strategy for REEVs supported by vehicle-to-everything (V2X) is proposed, where the two problems are coupled and a multi-input-multi-output (MIMO) reinforcement learning algorithm is proposed to simultaneously solve the two problems to obtain a cooperative strategy to achieve the optimal strategy when crossing signalized intersections. Also, a multi-objective reward function is designed by considering fuel economy and driving safety. The collaboration strategy results in improved fuel efficiency, battery wear, smoother speed profile for comfort, and engine operation near optimal BSFC for lower fuel consumption.