Thinking Tech

Video: Preventing deadly car accidents by leveraging the talent of Wayne Gretzky

Video: Preventing deadly car accidents by leveraging the talent of Wayne Gretzky

Posting in Environment

In the hopes of preventing deadly car accidents researchers have developed a new traffic simulation model that takes into account the unique quirks of human behavior.

We might think that traffic has a lot to do with the shape of a road, the turns, where it narrows, how many lanes, position of exit ramps, but there’s a subtle and elusive variable that makes the whole thing a lot more complicated. Human behavior.

Researchers have recently developed a computer simulation, SAFEPED, that includes not only geography and urban structures that might cause dangerous traffic accidents, but also the specific behaviors and psychology of drivers and pedestrians.

The researchers note that SAFEPED takes into account the quirks of human thinking that previous traffic models have not. They have analyzed how people make predictions and judgments while driving, and input those statistics into the model. The hope is to test new plans for intersections in the simulator, and then make adjustments to improve safety.

The simulator is based on a theory of visual perception developed by psychologist James Gibson. When we move through an environment we determine our so-called “optic flow,” taking into account our anticipated time of collision with another object or other people. For instance, the hockey player Wayne Gretsky had an incredible ability to determine optic flow when he’d pass the puck to a fellow teammate, knowing precisely when another player would be in a certain position at a given time to receive the puck.

In the SAFEPED simulator all virtual individuals think on their own, making decisions about their actions based on their specific visibility and their analysis of other drivers. So each individual’s action (eg., speed up, slow down, turn) depends on what they predict other drivers will do. Using the simulator scientists can view any accident from the point of view of any individual.  Then they can make changes to the virtual urban structure and press "play" again to watch what variables make real differences.

Already the simulation has provided interesting data.  If a person predicts that a car will cross an intersection in less than 2.5 seconds, they will not cross. But at 5.5 or 6 seconds most people will cross the road. The model also found that the further back a white stop line is set in an intersection, the smaller the chance that a pedestrian will be struck by a car.

Share this

Christie Nicholson

Contributing Writer

Christie Nicholson produces and hosts Scientific American's podcasts 60-Second Mind and 60-Second Science and is an on-air contributor for Slate, Babelgum, Scientific American, Discovery Channel and Science Channel. She has spoken at MIT/Stanford VLAB, SXSW Interactive, the National Science Foundation, the National Research Council, the Space Studies Board and Brookhaven National Laboratory. She holds degrees from the Columbia University Graduate School of Journalism and Dalhousie University in Canada. She is based in New York. Follow her on Twitter. Disclosure