Crowd Modelling for Quasi-real-time Feedback during Evacuation in a Situational Awareness System

Evacuation modelling has traditionally been limited to prior analysis of the geometrical design of spaces. This approach is limited by the number of scenarios that can be tested. A scenario-independent evacuation model would rely on real-time information processing. This approach allows the generation of an optimum evacuation strategy that minimises evacuation time and crowd density, and is tailored to the current situation at hand. This paper describes the requirements and development of such a crowd model. Graph theory and network flow heuristics are combined with agent movement, based on the individual determination of speed by meso-scale density assessment.