Using reinforcement learning to minimize taxi idle times

Abstract

Taxis spend a significant amount of time idle, searching for passengers. The routes vacant taxis should follow in order to minimize their idle times are hard to calculate; they depend on complex quantities like passenger demand, traffic conditions, and inter-taxi competition. Here we explore if reinforcement learning (RL) can be used for this purpose. Using real-world data to characterize passenger demand, we show RL-taxis indeed learn to how to reduce their idle time in many environments. In particular, a single RL-taxi operating in a population of regular taxis learns to out-perform its rivals by a significant margin.