During emergencies, lives are saved by the orderly, rapid evacuation of buildings. Efficient escape routes are important when architects plan the location of hallways and exits. Time is precious during actual emergencies, and nobody pays attention to the details. Fire drills can prove useful in improving evacuation times, but such tests are rather artificial. It would help architects if they could test how people would react in various layouts during a real panic, but it isn't ethical to fake such an event.
Ants may provide a way.
Civil engineers at Monash University in Australia set up a table-top building model with moveable walls and exits. They then populated the miniature structure with ants. After the ants had settled into their new habitat for a few days, the experiment began. Pellets of citronella ant repellant were inserted at various locations, and the resulting ant behavior was observed.
As the ants sought escape routes, observers noticed that the ants escaped twice as fast through openings in corners as through similar openings located midway along walls. Middle-of-the-wall exits tended to cause confusion and crowding of the ants. The ants escaped more efficiently through corner routes.