Scenario Evaluation of Autonomous Vehicle Spread Using Agent-Based Simulation

Abstract

This study analyzes the influence of autonomous vehicle (AV) on urban transportation using agent-based simulation. Assuming an AV share of 100%, the modes considered are private AV, car share AV and ride share AV. 10 AV spreading scenarios are defined based on AV use patterns and the dispatch algorithm in the case of sharing AV scenarios. MATSim which features co-evolutionary algorithms is used for the simulations. The target area is Numata City and Tone County in the northern part of Gunma Prefecture. The Gunma Prefecture 2015-2016 person-trip survey is used for calibrating the simulations. Findings suggest that even in rural cities highly dependent on cars, AV sharing can result in an increase in total travel distance, a reduction in the number of required vehicles, and relaxation of road congestion. Furthermore, while ride sharing might contribute to the reduction of environmental burden, excessive ride sharing might reduce the users’convenience considerably.