Critical phenomena, spontaneously arising or intentionally initiated in the crowd during mass events, can cause mass panic. If the surrounding area and escape routes are not prepared in advance, uncontrolled crowd behavior can produce crush and consequentially, causes casualties and injuries.
In our institute, we investigate the simulation of the behavior of large crowds of people in extreme situations based on the selection of scenarios that differ in the ways crowds move in a given territory. For this purpose, a multi-agent model is used, based on the formalism of social forces. Its parameters take into account the emotional state of the crowd, social structure, sex and age, as well as other external factors (e.g., weather). Based on the model, a procedure for short-term forecasting of the density and velocity of crowd movement in the territory is constructed. The interaction of individuals is also calculated, facilitating the assessment of the probability of injury or death.
This model can be used for planning preventive measures (overlapping streets, placement of fences and artificial objects, directing the movement of people flows) to ensure the safety of the event.
An example of the different scenarios of crowd panic developing in a rally in front of the Gostiny Dvor is illustrated in the video (St. Petersburg).