Crowd simulation technology is widely used in many aspects such as simulating the crowd evacuation during disasters, predicting emergency and so on. In this paper, the authors present a new model based on agent based model to simulate the crowd behavior under emergency situations in a confined space. Agents differ in aspect of whether having the knowledge of the topological structure of the confined space, which brings in the variance of evacuation routes. In this paper, authors proposed the conception of information asymmetry. Crowd combines the effect of the neighborhood and itself to generate the preferred velocity which indicates the motion trend. The solution process of preferred velocity is controlled by the neighborhood which is defined as the subjective factor, and the agent’s attributes which is treated as the objective factor. At the same time, a finite state machine is employed to realize the decision changing. The information asymmetry is necessary in crowd simulation under emergency situations according to the experiments in this paper.
