[2F5-GS-5] Agents:

[2F5-GS-5-01] Multi-Agent Simulation for the Uptake of New Mobility Services

〇Haruki Satoh¹, Fujio Toriumi¹ (1. The University of Tokyo)

Keywords:Mobility as a Service, Multi-Agent Simulation, Reinforcement Learning, Service Engineering

Most simulation studies on MaaS focus on the efficient management of the systems. However, considering the current situation where some MaaS projects have failed, the most important issue is to acquire users and monetize the business.
The purpose of this study is to explore scenarios where new services can acquire users against existing services. We constructed a multi-agent model consisting of two types of agents, services and users, and conducted simulations. As a result, we found that the availability of services is the most important factor in user acquisition, and that services specializing in a single characteristic are more likely to acquire many users. In addition, when services update their services, it tends to be more effective to reverse the trend of popular services than to imitate them. It was also suggested that initial campaigns by new services can greatly contribute to user acquisition.