A Behavior Based Crowd Simulation Framework for Riot Controlling in City Environment

Abstract

Creating complex and realistic crowd behaviors can be a difficult and time consuming task. By drawing the successful experiences of behavior-based AI and multi-agent simulation framework, this paper presents a behavior-based crowd simulation framework for riot controlling in city environment. A well-designed behavior-based prototype system is developed, which takes the advantages of the dynamics of interaction among the basis behaviors repertoires. The basis behaviors repertoires implemented by decision rules are composed of three parts, which correspond to the behaviors of the civilians, riots and soldiers respectively. Then these basis behaviors are combined into more complex behaviors by behavior selection mechanism. Eventually realistic crowd phenomena are created in the scenarios of riot controlling in city environment. We argue that this approach gives better results than conventional methods.