Detect the Formation of Groups in a Crowd Based on Social Relations

Abstract

People in public crowds tend to walk together based on their social relations, such as the relation-ships of friends, of family or colleagues, and to form groups. The works in [1,2,3,4] show that the percentage of people moving in a group is about 50% to 70%. This figure is much higher, above 70%, in stations or airports. In addition, individuals of these groups are usually connected with and rarely separated from each other thanks to their social relations. In this paper, we propose a model to detect the formation of groups in a crowd based on their social relations. Each group has a particular structure as well as certain roles of members in groups. The proposed model is experimented on the platform MASH [46]. The result of this study is used as a foundation to build a multi-level multi-agent model with the group level being a base level. An upper level is formed by the group representatives of its adjacent lower level and so on until the entire system is covered completely. We experiment with the model to simulate the behaviors of passengers in an airport lounge on fire. The simulated airport lounge has a structure like the lounge number 5, 1st floor, Danang International Airport. We approach the multi-agent system (MAS) to do this work.