PROLIFIC: Deep Reinforcement Learning for Efficient EV Fleet Scheduling and Charging

Electric vehicles (EVs) are becoming increasingly popular in ride-hailing services, but their slow charging speed negatively affects service efficiency. To address this challenge, we propose PROLIFIC, a deep reinforcement learning-based approach for efficient EV scheduling and services. The objective of PROLIFIC is to minimize passenger waiting time cost. formulates the problem as Markov decision process integrates distributed management model centralized order dispatching model. By using Q-network, agents can share supply information make interactions between dispatch decisions. This reduces curse dimensionality improves training efficiency neural network. proposed validated three typical scenarios with different spatiotemporal distribution characteristics order, results demonstrate that significantly cost all compared baseline algorithms.