Author(s)

S. Nakayama

Abstract

Transportation systems in general consist of many agents who choose their behaviors through learning based on their experiences and information provided. The agents interact mutually through the system, and the system must be dynamic and complex. The agent-based simulation is one of the methods for examining such a complex system. The simulation enables us to model the transportation system relatively flexibly. Assuming that agents reason and learn inductively based on their experiences, agent-based transportation system simulation models are developed. Each agent learns how best to choose a route based on his experiences, and the behavior of such agents and the mechanism of the transportation system are examined through simulation experiments. Keywords: agent-based simulation, transportation system analysis, network equilibrium, rule-based reasoning, route choice. 1 Introduction The behavior of a transportation system results from an aggregation of each individual’s travel behavior. The agent of travel behavior cannot predict the traffic state exactly before the trip, but obtains ex post facto information such as how many minutes it costs after the trip. So, he learns how to choose a route, departure time, or other travel choice based on his experiences. This may include prediction of traffic state before the trip. The state of the transportation system is given by an aggregation of these travel behaviors. The agents interact mutually through the traffic state, recording travel times. We cannot appreciate or understand the mechanism of the system that includes nonlinear interaction by analyzing parts or elements of the system separately [1]. We have to investigate the whole system simultaneously. This

Keywords

agent-based simulation, transportation system analysis, network equilibrium, rule-based reasoning, route choice.