Optimizing Automated Mobility on-Demand Operation with Markovian Model: A Case Study of the Tel-Aviv Metropolis in 2040

Abstract

The emergence of Autonomous Mobility on Demand (AMoD) services heralds a transformative shift in urban transportation dynamics. With their potential to significantly reduce operational costs and eliminate the need for drivers, AMoD services are poised to revolutionize mobility in cities like Tel-Aviv, where they are projected to capture a substantial portion of daily trip demand. Despite their promise, comprehensive studies evaluating the full functionality of AMoD services, especially charging behavior under real-world conditions, remain scarce. Following this gap, our study delves into the core tasks of AMoD fleet management: dispatching, routing, charging, and rebalancing. Leveraging advanced simulation tools, we undertake a rigorous examination of AMoD operations to predict demand, enact daily plans, and optimize fleet activities through a sophisticated Markov decision process (MDP) model. Our findings reveal that the MDP model facilitates the derivation of optimal actions for individual AMoD vehicles, thereby maximizing future profitability while fostering substantial energy savings.