A novel eco-driving strategy for heterogeneous vehicle platooning with risk prediction and deep reinforcement learning

-Eco-driving control holds significant potential for energy savings in vehicle platooning. However, the development of energy-saving strategies is hampered by complex traffic scenarios and the heterogeneity of vehicles within the platoon. This study proposes a novel eco-driving approach that comprehensively accounts for the dynamic uncertainty of obstacles in avoidance scenarios and the heterogeneity between the lead vehicle and follow-up vehicles. First, the dynamic and powertrain models for the follow-up vehicles are constructed, and a rule-based risk prediction model is designed to predict the risk level during platoon obstacle avoidance. Besides, a deep reinforcement learning-based controller is developed to optimize longitudinal speed, balancing objectives such as driving efficiency, safety, and energy consumption, thereby reducing hydrogen consumption while ensuring driving safety and driving efficiency. The simulation results indicate that the proposed eco-driving approach eliminates the impact of heterogeneity within the platoon, achieving hydrogen consumption savings of 1.47 %-4.79 % under complex obstacle avoidance scenarios.