Improvement and evaluation of an evacuation behavior simulation model

We constructed a multi-agent simulation model to replicate the results of an evacuation study in a field experiment [4]. Four or two leaders instructed 16 evacuees in the follow-me method. The results of the simulation, however, were not close to the results obtained in the experimental study. Therefore, we introduced a rule that evacuees follow a leader if the evacuee happens to see a leader instructing another evacuee. In this case, with four leaders, the results obtained were in good agreement with the experimental ones. With two leaders, however, the agreement between the simulation and experimental results was still poor. According to another study [9], there are two types of evacuees. One type actively follows the leader instructions, whereas the other type follows the crowd movement. We suggest that, in the case of two leaders, better agreement would be obtained by reflecting the two types of evacuees in the simulation model.