Modeling hallway traffic and panicked stampedes

Science : Current crowd simulations, used to develop safe public spaces, usually treat virtual humans within the model as simple particles that repel each other and zoom to their destinations. While these models are useful for predicting how long pedestrians might take to cross an intersection, they cannot reconstruct or predict the chaotic movement of a large number of people trying to escape a crowded room. Mehdi Moussaid, of the University of Toulouse, France, and colleagues created a computer model that emphasizes human behavior rather than many-body physics. The model doesn’t ignore physical laws, such as the person-to-person energy transfer that help explain “crowd-quakes”, but it does rely on heuristic formulas the team derived from studying patterns of pedestrian movement in videos. Moussaid’s approach is the first to accurately reproduce pedestrian behavior across a spectrum of intensity: from lanes of pedestrians formed in simulations of hallway interactions to emergency escapes from bottlenecked rooms.