Game Theoretic Considerations for Optimizing Efficiency of Taxi Systems

Abstract

Taxi service is an indispensable part of public transport in modern cities due to its unique end-to-end convenience, flexible timetable, and comfortable riding experience. To support these advantages, taxi systems adopt a decentralized operation mode where taxi drivers strategically decide their schedules and routes and compete with each other for individual profits regardless of the system level efficiency. Not surprisingly, taxi systems are therefore usually inefficient and difficult to optimize. While taxi drivers’ strategic behavior plays a central role here, they were unfortunately ignored in most existing research. To overcome the inadequacy, we proposed a game-theoretic approach to model taxi drivers’ behavior. We apply the model to optimizing the pricing scheme of taxi markets, and in particular to solving a long-standing issue known as the peak-time dilemma in Beijing's taxi market.