NERO: Nested Rebalancing Optimization for Mobility on Demand

Abstract

Mobility-on-Demand (MoD) services, such as taxi-like services, are promising applications. Rebalancing the vehicle locations against customer requests is a key challenge in the services because imbalance between the two worsens service quality (e.g., longer waiting times). Previous work would be hard to apply to large-scale MoD services because of the computational complexity. In this study, we develop a scalable approach to optimize rebalancing policy in stages from coarse regions to fine regions hierarchically. We prove that the complexity of our method decreases exponentially with increasing number of layers, while the error is bounded. We numerically confirmed that the method reduces computational time by increasing layers with a little extra travel time using a real-world taxi trip dataset.