Agent-based research on crowd interaction in emergency evacuation

With the rapid development of economy, large-scale public buildings have been constructed in many countries. However, if emergency such as fire appears, it is very difficult to escape from such kind of structure where a huge crowd gathers. Therefore, how to evacuate safely, effectively and fast becomes a very hot research topic in recent years. In fact, most of the scholars focus on the analysis of personal thought, behavior and psychological characteristics, neglect the crowd interaction. First, we define every person to an agent. The crowd consists of three kinds of agent: Lead Agent, Ordinary Agent and Panic Agent. Second, the mechanism of crowd interaction is analyzed. Third, a crowd interaction model based on agent and cellular automaton is developed. Finally, the Harbin subway station is selected as simulation object, adopts software Building EXODUS to simulate crowd interaction. Results show that crowd interaction especially the Lead Agent's guiding and coordination can help to improve the evacuation efficiency and shorten evacuation time.