In 2008, 762 fatalities in the U.S. resulted from drivers running red lights at intersections, according to the Federal Highway Administration. Each year, an estimated 165,000 people in the U.S. are injured by red light runners.
To reduce such accidents, researchers at MIT have come up with an algorithm that flags oncoming cars as likely red light runners a few seconds in advance. The algorithm’s calculations are based on parameters like the vehicle’s deceleration and its distance from a traffic light.
It correctly identified red light violators 85 percent of the time. Perhaps more importantly, the algorithm generated fewer false positives (i.e. flagging compliant vehicles as likely red light runners) compared to similar prediction technology.
The MIT scientists hope to use this algorithm to allow “smart” cars of the future to alert drivers.
At intersections, the algorithm would tell a driver to not go ahead, even when the light is green, if it detects a likely red light runner.
Not generating too many false positives, therefore, is important.
“If you’re too pessimistic, you start reporting there’s a problem when there really isn’t, and then very rapidly, the human’s going to push a button that turns this thing off,” said Jonathan How, an MIT professor involved in the development of the algorithm.
Implementing such a warning system requires vehicle-to-vehicle (V2V) wireless communication, according to How.
An MIT press release regarding this algorithm stated that the U.S. Department of Transportation is exploring V2V technology with major car manufacturers like Ford, which is already road-testing prototype vehicles.
The MIT researchers are also planning to adopt this algorithm to air traffic control.