知的マルチエージェント交通流シミュレータMATESの開発  [in Japanese] Development of Multi-Agent Traffic and Environmental Simulator MATES  [in Japanese]

This paper describes a development of an advanced traffic simulator based on a multi-agent approach which is named MATES Multi-Agent Traffic and Environmental Simulator. In this research, city traffic phenomena are basically regarded as complex systems consisting of a number of human beings. Each car driver is modeled as an intelligent agent with its own logic and preference. Various different types of characteristics of drivers can be easily implemented into the agents. The environment surrounding each car consists of other cars, road network, traffic signals, pedestrians and so on. Interaction among numerous agents in the environment generates nonlinear behaviors of city traffic phenomena. After describing details of the theory and implementation of the MATES, its fundamental features are first examined. Next the MATES is applied to simulate city traffic in KASHIWA city in Chiba, employing various real world data as input. The simulation results are compared with monitored ones. Through those simulations, it is clearly demonstrated that the MATES is a powerful tool to study complex city traffic problems precisely.