Chinese Team Develops Multi-Objective Deep Reinforcement Learning Strategy for Safer, More Efficient Autonomous EVs

Chinese researchers have recently proposed a novel autonomous driving strategy called "Intelligent Energy-saving Driving Strategy" (IEDS), designed to simultaneously enhance the safety and energy efficiency of plug-in hybrid electric vehicles (PHEVs). Built upon deep reinforcement learning with a multi-head Deep Q-Network (DQN), this strategy processes real-time data on vehicle status, traffic conditions, and road environments to jointly determine lane-changing maneuvers, speed adjustments, and torque distribution between the internal combustion engine and electric motor. Unlike conventional approaches that treat safe and eco-friendly trajectory planning and energy management as separate tasks, IEDS achieves end-to-end collaborative optimization, balancing safety and energy savings in complex driving scenarios. This advancement provides key technical support for future green and intelligent mobility. The findings have been published in the journal *Green Energy and Intelligent Transportation*.