Agent driving behavior modeling for traffic simulation and emergency decision support

Traffic evacuation is one of the most important tasks in emergency management, and it is challenging for governments to plan an efficient and safe evacuation before a huge disaster strikes. This paper presents a traffic evacuation simulation system that generates agent´s driving behavior based on multi-level driving decision models. The agent´s driving behavior combines multiple widely used behavior models from each decision level. The agent-based traffic evacuation system is proposed and a prototype system implements each agent´s multi-level modular driving decision models. The simulation experiment studies show varied clearance time, evacuation rates per shelter, and the variety and number of traffic jams to support traffic evacuation planning decisions in a crowded city liked Beijing, China. The simulation studies compare the existing evacuation plan with other simulated plans and evaluate it for designing a better and more realistic traffic evacuation plan.