Agent Guidance in Autonomous Mobility on Demand Systems: An Approach Utilizing Priority Double Deep-Q-Networks

Autonomous Mobility-on-Demand (AMoD) systems have become an integral part of modern urban life, reshaping transportation dynamics. However, the challenge of orchestrating multiple vehicles to maintain a dynamic equilibrium between supply and demand in a multi-agent environment remains a critical concern. To address this challenge, we propose a novel Multiagent Deep Reinforcement Learning framework called Priority Double Deep-Q-Network (Pr-DDQN) with a new cooperative reward mechanism to optimize the repositioning route of vacant vehicles within the AMoD system. Through rigorous experimentation using a city-scale dataset comprising 48,000 requests on a weekday in Chicago, we assess the scalability and efficiency of our approach. Comparative results demonstrate that Pr-DDQN outperforms existing methods, showcasing superior performance across key metrics, including Service Rate, Satisfaction Index, and Repositioning Time. These findings underscore the efficacy of our approach in enhancing the operational efficiency and overall performance of AMoD systems.