MIT Algorithm Can Predict Red Light Runners

MIT Algorithm Can Predict Red Light Runners

Statistics gathered by the National Highway Traffic Safety Administration reveal that 700 of 7,000 traffic related fatalities in 2008 were caused by drivers that failed to stop at red traffic lights. What’s more, the red light offenders and the victims involved in the accident (whether passengers, pedestrians, or the driver of the other vehicle) share a near equal, 50/50, chance of fatality.

Determined to eliminate this danger, researchers at Massachusetts Institute of Technology developed an algorithm used to predict the likelihood of an oncoming vehicle of running a red traffic light. According to MIT’s aeronautics and astronautics professor, Jonathan How, the growing multitude of active safety advanced telematics systems could benefit from algorithms that prevent drivers from getting into accidents.

The algorithm was developed by Georges Aoude, a former student of How, and applies parameters that can rate a vehicle’s rate of deceleration and distance from the traffic light at intersections to determine whether a vehicle is likely to run the light. If this algorithm becomes widely accepted by multiple automakers, calculated data could be communicated from vehicle-to-vehicle. Professor How explains, “If you had some type of heads-up display for the driver, it might be something where the algorithms are analyzing and saying, ‘We’re concerned.’ Even though your light might be green, it may recommend you not go, because there are people behaving badly that you may not be aware of.”

A field test for the algorithm was conducted at an intersection in Virginia, and from the recorded results of 15,000 approaching vehicles, Georges Aoude’s algorithm achieved a hit rate of 85 percent. The determination of  a red light offender is calculated a couple of seconds before the vehicle reaches the intersection, which researchers believed are sufficient to allow other drivers to react.

The research team is also developing the generation of a recommended response for the driver in the event that danger in the intersection is sensed. What’s more, there are plans to tweak the algorithm for the application of air traffic control.

[Source: New York Times]

  • Christopher F. Lussier ’97

    You do not need an algorithm to figure this out. I currently live in north Long Beach, California. This borders the city of Compton. Without trying to be or act or generalize any type of predjudice, I hereby point out that we have a very high population of inner-city black people. This particular segment of the population live out their lives with a “ghetto mentallity”. This means, they run around thinking the world owes them a living and have a general poor attitude towards everybody and anybody. This is most prevalent when they are operating a motor vehicle. They speed, They pass on the right outrunning the cars stopped for a red light, just as the light turns green. They deliberately accelerate on a yellow light just so they do not get caught having to wait for a light. Most of the time they are seen crossing in front of you at 45 mph after you already have your green light. The solution is very simple. And, no, red light cameras are not the answer as most have someone elses plate on their vehicles. Here is the cure. Simply program a three or four second delay into the timing of a light controlled intersection. Place the delay right between when one direction gets a red light and before the cross traffic gets their green light. Its that simple. Why I do not see this in the programming of the lights anywhere I look just baffles me.