Developing an Agent-Based Simulator Combining Mesoscopic Traffic Simulator with Dynamic Vehicle Allocation System to Evaluate a Ride-Sharing Service in Urban Area

In this study, we developed an advanced agent-based simulator which allows an agent-based mesoscopic traffic simulator to cooperate with a dynamic vehicle allocation system. This simulator can reproduce a dynamic vehicle allocation service in real world through its Web API. We implemented this simulator to evaluate ride-sharing taxi services in Kumamoto City to compare with the conventional taxi services. Based on simulation results, the ride-sharing taxi service is more effective. The average time from booking a ride to arriving at the intended destination was significantly reduced as the number of vehicles increased; however, the average occupancy rate of vehicles decreased.