Fire Safety Science Digital Archive

The research of human crowd dynamic is significant for the building designers and emergency guiders. However, current researches mainly focus on the interactions among isolated individuals, the influence of social interactions among pedestrian is needed to take into consideration in the pedestrian dynamic study. In this manuscript we carried out observed experiments of the movement of pedestrian group and used optical flow algorithm to extract pedestrian’s trajectory from the video. Then the movement characteristics of the group members were analyzed and discussed, which include the angle, the distance, the velocity, the group alignment, step frequency and the offset angle of group member. It indicates group members have adaptive ability to maintain the group structure and the velocity fluctuation of the group is small. It is found that the velocity and step frequency of group member are smaller than that of individuals under the same scenario. Movement consistency of the group members is also verified. These experimental results will be a support for modeling pedestrian group.