A Multi-agent Simulation of Traffic System under Mixed Traffic Condition

Traffic system is difficult to be simulated because of the emergence phenomenon. Traditional simulation research has achieved great development, but most of them are based on ideal conditions which assume vehicles have little difference in dimension, maneuverability or other aspect. In China mixed traffic occupies the majority, so big errors might occur when mathematical method is used. Multi-agent technology is suitable for such a complex simulation problem. In this paper, multi-agent method is used for typical mixed traffic microscopic simulation. The characteristics of mixed traffic are expressed in two ways: 1) different kinds of vehicle-agent models with different maximum-velocities, 2) fuzzy logic of driver agent that explicitly represents drivers´ decision-making process of acceleration. To verify the effectiveness of our method, we compare the simulation result with real traffic.