Page 3.14

Virtual Crowding

When does a large crowd become a dangerous mob? Where should architects put emergency exits to best anticipate how people will react under sudden fear? It’s hard for scientists to answer these questions, mostly because the environmental situations that incite crowd behaviors can’t be simulated in real life. As Arizona State University geographer Paul Torrens explains: “You couldn’t stage a realistic rehearsal of an evacuation because people are not going to panic appropriately.”

One way to get around this is to model the situations in virtual reality. Below are screenshots of scenes produced by Torrens’ geosimulation lab, whose programs integrate individual “agent” characteristics—like age, sex, and body language—with group behaviors and environmental variables.

In this screenshot, the crowd flees a burning car, toward an evacuation point on the right-hand-side of the screen. Narrow streets form a bottleneck that constrains the densely-packed crowd.

The simulation can be run many times over with varying conditions and from varying vantage points.

Synthetic pedestrians are texture-mapped with realistic clothing and features. The model makes use of motion capture data to lend individual gait and body language to each pedestrian.

To read more about Dr. Torrens’ geosimulation research, visit his lab’s website.

Image and Caption Credits: Paul Torrens, Arizona State University