Data-Driven Modeling on Spatial Pattern and Private Space of Pedestrian Groups

This paper studies spatial patterns and private spaces of pedestrian groups. We found that group members always maintain relatively fixed positions when walking, so that the group keeps stable patterns in low density free pedestrian flow. Through empirical observations and statistics, common spatial patterns of groups with different numbers of members were derived. By analyzing the relationship between the interactive behavior of group members and of group spatial patterns, we defined several functional zones of the group and established the concept of Expected Private Space of Group (EPSG). On the basis of calculating distribution of group members around the barycenter with observational data, we obtained the shape and location of each functional zone, and a geometric model of EPSG was built to describe the formation variation of the groups in a pedestrian stream.