Multi-layered framework for crowd microscopic simulation

We address the challenging open problem of simulating virtual pedestrian crowd behaviors in real time inside virtual environment, in this context; we propose an overall architecture for defining the natural structure of crowd at three levels, the individual agent, group behaviors, and the relationships among individual agents and groups. Each level has its own complexity. The purpose of this paper is to create a model that try to cover all or most natural behaviors, so to do this, we propose a microscopic solution based on multi-layered architecture consisting of three level which together model the emergent crowd behaviors; the first layer, the physical layer realizes the perception and motion processes, the behavioral layer formulates the pedestrian behavior in order to follow his path together a goal, and finally the navigational layer which is responsible to implement the autonomous navigation process.