Evolution of pedestrian evacuation considering different human behaviors

Pedestrian evacuation becomes more complicated when different human behaviors are considered. In this paper, pedestrian evacuation considering two different kinds of human behaviors is studied by using a cellular automaton (CA) model. Two types of evacuees, cooperators and defectors, are initially randomly distributed in a square room with one exit and they interact with their neighbors following the snowdrift game theory (characterized by the parameter “r”) during the evacuation process. Cooperators and defectors are mutually changed according to two typically different kinds of evacuees’ judgements, namely, judgement based on neighbors’ strategies (we call it herding behavior) and judgement based on neighbors’ payoffs (we call it independent behavior), which leads to the corresponding evacuee clusters. It is found from a large amount of numerical simulations that the proportions of the corresponding evacuee clusters are evolved to respectively consistent states despite 11 typically different initial evacuee cluster ratios, largely owing to self-organization effect. Moreover, for other 200 randomly chosen initial evacuee cluster ratios, Monte-Carlo simulations show that the proportions of the corresponding evacuee clusters are evolved stably, without any oscillation, and when the initial number of pedestrians of two different kinds of behaviors are comparable, the proportions are evolved to respectively consistent states despite different initial evacuee cluster ratios, indicating that some regular phenomena occur during the evacuation process. Furthermore, it is found that the moderate values of r is necessary for short evacuation time, and for the fixed values of r, the initial ratio of pedestrians of herding behavior has a great influence on the final evacuation time.