Animal dynamics based approach for modeling pedestrian crowd egress under panic conditions

Collective movement is important during emergencies such as natural disasters or terrorist attacks, when rapid egress is essential for escape. Collective patterns are not restricted to humans, but have been observed in other non-human biological systems. Social animals often move in groups: ant trails, wildebeest migrations, and locust swarms are some of the natural phenomena at the heart of a movement ecology (Holden, 2006). Although most collective movements are routine, there are rare but perilous crowd panics that may sharply affect survival and fitness, like stampedes of wildebeest under predator attack, evacuations of ant nests in the face of flooding, or flights of people from burning buildings. Not enough is known about the underlying dynamics of crowd panics, despite the obvious importance of such knowledge in the human sphere. Studies of human crowds during evacuations were carried out as early as 1930s (Kholshevnikov and Samoshin, 2008), but the problem of understanding panicking groups and enhancing safety under emergency conditions still exists (Helbing et al., 2000).The complex interactions of panicking individuals with their social and physical environments make theoretical understanding difficult. Models of pedestrian behavior in emergency situations rarely have complementary empirical data to validate the model s prediction, so we may not want to rely entirely on mathematical models before scaling up to an applied, real world situation. That lack of data is most likely a major factor explaining why very few models exist which focus on panic situations. The bulk of the literature is restricted to the study of normal (non-panic) pedestrian dynamics or normal evacuation processes (Still, 2000; Hoogendoorn and Bovy, 2002; Hoogendoorn, 2004; Daamen, 2004; Antonini et al., 2006; Kretz, 2007; Asano et al., 2009). Even the researchers who developed the few existing models of crowd panic have identified the need for more rigorous modeling frameworks and the development of approaches to assess the reliability of model predictions.