Game theory serves as a good tool to explore crowd dynamic conflicts during evacuation
processes. The purpose of this study is to simulate the complicated interaction behavior
among the conflicting pedestrians in an evacuation flow. Two types of pedestrians, namely,
defectors and cooperators, are considered, and two important factors including fear index
and cost coefficient are taken into account. By combining the snowdrift game theory with
a cellular automaton (CA) model, it is shown that the increase of fear index and cost
coefficient will lengthen the evacuation time, which is more apparent for large values
of cost coefficient. Meanwhile, it is found that the defectors to cooperators ratio could
always tend to consistent states despite different values of parameters, largely owing to
self-organization effects.