Using physics to prevent human stampedes

Brett Smith for – Your Universe Online

Walking through Mid-town Manhattan can feel a bit like the mosh pit at Foo Fighters concert, with less of the violent body-to-body contact. Researchers have long been fascinated by the dynamics of this kind of crowd movement, and science has struggled to produce a model that can reliably replicate crowd dynamics.

Now, for the first time, scientists have developed a simulation that can reliably model crowd dynamics, according to research being presented on Friday at the American Physical Society’s March Meeting.

To create the model, the researchers started by considering individual people in a crowd as a bunch of charged particles. If humans do behave like charged particles, the model could be based on a simple “repulsive force” that acted across a certain distance. However, this model doesn’t reliably reproduce crowd dynamics.

[STORY: Most people would rather do anything at all than sit in silence]

The team needed to develop a more complex model and to do so – they analyzed video footage of crowds in both an outside campus setting and an indoor setting involving a bottleneck. The researchers learned that people engage in a reliable and universal way if their “time until a possible collision” is known. The study team found that humans do not act like charged particles since they can foresee collisions, and this is crucial to outlining crowd interactions.

Time Until Collision would be a cool band name

Based on their investigation, the scientists discovered a simple mathematical law for the interactions force between two pedestrians that relies only on the “time until collision” factor. They used this rule to emulate realistic crowds in an array of urban settings.

When they ran their simulation, the team saw familiar crowd motions emerge: people strolling towards an on-coming collision veered well ahead of time, and individuals traveling in exactly the same direction usually walked near each other. These real-life patterns had been tough to reproduce with previous pedestrian models.

“Remarkably, this simple law is able to describe human interactions across a wide variety of situations, speeds, and densities,” the study team wrote about their work in an abstract published by the APS.

The researchers noted that their research united the fields of statistical physics and sociology.

“It is very exciting to me to think about how psychology affects our motion and how people are different from the systems physicists normally model,” said study team member Stephen Guy, a 3D modeling and animation expert at the University of Minnesota.

The study team said they want their crowd simulation to be used to anticipate possible dangers in sporting events and music festivals – possibly preventing congestion and stampedes. Their models could also be used to test the crowd safety of new structures prior to any construction occurring.

[STORY: ER visits on the rise due to distracted walking]

“The universality of our law is really surprising, and understanding this can lead to safer building designs and shed some light into the anticipatory nature of human interactions,” said study author Ioannis Karamouzas, a post-doctoral engineering researcher at the University of Minnesota.

Karamouzas specializes in computing for virtual characters and the new research he is scheduled to present could also be applied to more realistic crowds in video games and other kinds of digital entertainment.


Follow redOrbit on Twitter, Facebook, Google+, Instagram and Pinterest.