[4N3-J-7] Agents: mobility of humans and vehicles

[4N3-J-7-03] Study of Adaptive Hybrid Traffic Simulation

〇SHIHO ONO¹, Hideki Fujii¹, Shinobu Yoshimura¹ (1. University of Tokyo)

Keywords:multi-agent simulation, hybrid traffic simulation

In this research, an adaptive approach for hybridization of microscopic and mesoscopic traffic simulation based on multi-agent system is focused on. In a microscopic model, a car agent acts according to the result of interactions with other agents. In a mesoscopic model, a car agent acts according to macroscopic variables such as traffic density, traffic capacity and free flow speed of road segment. Calculation speed is expected to be enhanced by using a light-weight mesoscopic model in the area with low traffic. At the same time, fine granularity near intersections with heavy traffic can be expressed by using a precise microscopic model. We investigated the relationship between inflow traffic volume and traffic flow characteristics of each model as the first step. The result shows that the difference of two models is increased under the condition that right-turn cars frequently give over the right-of-way to oncoming cars.