Simulation based dynamic traffic assignment (DTA) models have received much attention in recently years. Compared to static traffic assignment models, DTA models consider time variations in the demand and flow behaviors in the traffic network. However, within the current DTA framework, it is difficult to model traveler’s characteristics and behaviors such as the heterogeneity, information sharing, learning process and the interactions among the travelers. Agent-based modeling and simulation (ABMS) is specifically developed to address this complexity and to support individual agent-based decision making. This paper proposes a framework that integrates an agent-based joint route and departure time choice model with the DynusT traffic simulation. The proposed agent-based joint route and departure time choice model considers learning from previous experiences (e.g. trips), heterogeneity of different travelers, incomplete network information, and information sharing between travelers. An interface is defined between the simulation package and the proposed choice model that supports individual agent to make route and departure time decision jointly. The generalized cost consists of arrival time penalty and travel time cost serves as the criterion for choosing departure time and route. The results of the case study indicate that travelers tend to choose the route and departure time with the lowest generalized cost.

DynusT;Agent-Based Model;Travel Behavior;Intrgration;Arrival time penalty function