A Distributionally Robust Reinforcement Learning Framework for Constrained Urban EV Dispatch
Researchers have developed a new framework for controlling urban electric vehicle (EV) fleets that uses distributionally robust reinforcement learning (DRL) to handle uncertain demand and travel times. This approach, called PD-RSAC, optimizes dispatch, repositioning, and charging decisions while strictly adhering to charger and feeder capacity limits. Experiments using New York City taxi data demonstrated that PD-RSAC significantly increased net profit to $1.22 million, outperforming various heuristic and reinforcement learning baselines. AI
IMPACT This framework could improve the efficiency and profitability of large-scale urban mobility services by optimizing EV fleet operations under uncertainty.