An Agent-Based Simulation Model for Shared Autonomous Taxi System

Abstract

The shared autonomous taxis system (SATS) has been regarded as a promising traffic mode for improving travel flexibility and reducing travel costs. This study aims to examine the potential benefits of replacing all taxis with ride-sharing autonomous vehicles (AVs). Specifically, two sharing strategies are discussed: nondetour sharing, in which a subsequent customer is picked up only if no detour is required, and detour sharing, where the detour may cause a delay for the first customer. An agent-based simulation is developed to demonstrate the advantage of the SATS. Results show that the nondetour and detour sharing strategies can respectively reduce fleet size by 19% and 27%, reduce waiting time by 62% and 82%, reduce operational costs by 16% and 24%, and reduce CO2 emissions by 17% and 19% in comparison with a nonsharing strategy.