To save lives, predicting which cars are most likely to run red lights

MIT researchers have developed an algorithm that can accurately predict when a car is likely to run a red traffic light. With V2V technology, it's the latest work in the development of a "smart" car.
Written by Andrew Nusca, Contributor on

Last night, like any other, I took my dog for a walk. And last night, like many others, I was nearly run over by a driver who could not resist the urge to roll through a stop sign.

It's a major problem in my area of the city, a residential neighborhood where stop signs, not traffic lights, rule. But the problem exists at intersections with lights, too.

The British solution to this problem is the roundabout, which has been proven to reduce the number of auto collisions. For some reason the technique hasn't caught on in the U.S. -- okay, except maybe for the New Jersey "jughandle" -- but the problem persists.

It's not just drivers who are to blame. We've all seen a brazen bicyclist disregard traffic signals, or a hurried pedestrian run into traffic. It's the 21st century, we're all in a rush, and we're all risking our lives to save a few seconds.

In order to reduce the number of fatal collisions at intersections, researchers at the Massachusetts Institute of Technology have developed an algorithm that they say predicts when an oncoming car is likely to run a red light.

The algorithm uses parameters such as the vehicle’s deceleration and its distance from a light to determine which cars were potential threats. Or as they like to call them, "violators."

The researchers tested the algorithm on data collected from an intersection in Virginia and found that it accurately identified potential violators 85 percent of the time, within a couple of seconds of reaching a red light -- enough time, they believe, for other drivers to react to the threat if alerted.

The hurdle is figuring out how to actually convey those alerts, which are delivered in just five milliseconds. The hope is that the rise of the "smart," connected car will allow vehicles to communicate with each other and the infrastructure around them to warn drivers of a potential dangerous situation -- and, given the intelligence, could even stop on their own.

It's called V2V technology, and it's under development by major automakers. But once the infrastructure is in place, algorithms will be needed to make such communication useful. The MIT researchers are part of the solution, and are even exploring how their algorithm could be applied to air traffic control situations for airplanes.

Their research will appear in the journal IEEE Transactions on Intelligent Transportation Systems.

Related on SmartPlanet:

This post was originally published on Smartplanet.com

Editorial standards