Using non-human biological entities to understand pedestrian crowd behaviour under emergency conditions

• We test a single modelling framework to examine crowd behaviour in ants and humans.
• The predictions of model correspond to the empirical data available for panics.
• Obstacle near exit and homogeneity of crowd have significant effect on outflow.
• Broader comparison of crowd behaviour among species can provide insight on crowd panic.

Models of collective movement have been developed for both human crowds and animal herds and other aggregations but these models have not been used to test whether panicked crowds display generic features of dynamical behaviour regardless of species, and in particular whether a single model can explain panic behaviour in organisms of vastly different body size. We use a single modelling framework to examine crowd behaviour in ants and humans, which differ by 8 orders of magnitude in body mass. We assess whether simple allometric scaling of model parameter values, based only on the body mass difference, allows the model to describe the collective behaviour of the two species under panic conditions. We verified the model against experimental data from panicking Argentine ants (Linepithema humile) and then rescaled the parameter values to human body size. The predictions of rescaled model correspond to the quantitative data available for crowd panics, suggesting that the same kinds of interactions among individuals and with the physical environment govern crowd behaviour. We tested the effects of partial obstruction and the homogeneity of body sizes on the escape rate and found that appropriate selection of size and location of obstruction and homogeneity of body sizes can increase the outflow of pedestrians by more than double. Broader comparisons of crowd behaviour among species with different forms of locomotion and body size can enhance our theoretical understanding of crowd panics and potentially has applications in handling of agricultural animals as well as human public safety.