关键词: 城市交通, 行人仿真, 社会力模型, 同伴群, 自组织现象

Abstract:

Walking together groups, such as family, friends, etc., often occupy a high proportion in pedestrian crowd. Group members tend to walk side by side or form a shallow V-shape pattern in order to communicate each other. Among previous studies, social force based group pedestrian simulation model described it best. However, our observation notices that when density of crowd or number of group members increases, the structure turns into a dynamic sub-group based multi-row pattern. This self-organized pattern emerges from the movements’combination of individual group members, who seek for a proper position to satisfy their communication need as well as to reduce the marching resistance from other pedestrian or surroundings. We introduce a dynamic sub-group forming model to reflect the decision making process of group members. Furthermore, simulation experiments are conducted to analyses how group behavior affect the crowd dynamics. The experimental results show that: this model can replicate the dynamic sub- group based multi-row pattern when the density of pedestrian is high or the group members is large; the existence of groups in crowd decreases the speed of pedestrian flow but the effect weakens with the increase of flow density; our model can improve the efficient of simulated pedestrian flow and make it closer to the actual crowd dynamics.

Key words: urban traffic, pedestrian simulation, social force model, social group, self-organization