Evacuation during Violent Attacks: Agent-Based Modeling and Simulation

In public open space there are often many civilians engaged in normal activities. Violent attack against civilians may lead to a large number of casualties. There is an urgent demand for a tool which could systematically explore the evacuation process of crowd under violent attacks. This paper presents the general characteristics of public open space and build an agent-based model of evacuation under violent attack. All the participants are classified into two types of agent, the civilians and the attackers, and there is complex interaction among them. First, the activity of civilians and attacker are described with a framework of finite state machine. Then, focusing on the attacker using a sharp instrument, this paper analyzes his target search strategy, and quantitatively model the killing range of sharp instruments. For civilians, their exit choice behavior is formulated based on discrete choice model, and the civilian's movement is calculated with an expanded social force model. Finally, the simulation platform is developed based on Anylogic. The simulation results of a typical scenario show that the platform could be used to study evacuation dynamics of crowd during violent attacks.