Decision time? Check out our latest product comparisons

Algorithm predicts which cars are most likely to run red lights

By

December 1, 2011

MIT has developed an algorithm that predicts which cars are likely to run a red light, so ...

MIT has developed an algorithm that predicts which cars are likely to run a red light, so that other drivers can be warned

According to the U.S. National Highway Traffic Safety Administration, for the year 2008, over 700 fatalities resulted from drivers running red lights at intersections across the United States. Approximately half of the people killed weren't the errant drivers themselves, but were other drivers, passengers or pedestrians who simply happened to be in the wrong place at the wrong time. One approach to reducing these numbers is to utilize technology such as Mercedes Benz's Smart Stop system, that won't let drivers run red lights. Scientists at MIT are looking at the problem from another angle, however - they have developed a system that identifies cars likely to run the reds, so that the other drivers can be warned to stay out of their way.

The system uses cameras to view vehicles heading towards an intersection, then applies a custom algorithm to parameters such as their rate of deceleration, and distance from the light. That algorithm was tested using data previously gathered at a busy intersection in Christianburg, Virginia. Out of over 15,000 approaching vehicles within the data-set, the system correctly predicted which ones would run red lights with 85 percent accuracy.

The system was particularly accurate within a window of one to two seconds before a collision could potentially occur. While that might seem like the equivalent of forecasting rain right as the dark clouds are rolling in, the researchers believe that even those couple of seconds would provide a sufficient warning time for other drivers approaching the intersection.

That warning would likely take the form of an alert on a vehicle-to-vehicle communication system, sent by the offending car to all of those in its vicinity.

While the MIT system is not the first of its kind, it is reportedly 15 to 20 percent more accurate than any that have come before it. In particular, it raises fewer false alarms, which would make it less annoying for drivers to use.

The research team is now looking into getting the system not just to alert motorists to the presence of red-light-runners, but to suggest evasive actions. There are also plans to adapt the algorithm for use in air traffic control, where it would predict what aircraft were likely to do.

About the Author
Ben Coxworth An experienced freelance writer, videographer and television producer, Ben's interest in all forms of innovation is particularly fanatical when it comes to human-powered transportation, film-making gear, environmentally-friendly technologies and anything that's designed to go underwater. He lives in Edmonton, Alberta, where he spends a lot of time going over the handlebars of his mountain bike, hanging out in off-leash parks, and wishing the Pacific Ocean wasn't so far away.   All articles by Ben Coxworth
Tags
3 Comments

I can just imagine this, that arrogant driver saying "it's ok for me to run red lights, my vehicle to vehicle system will warn other drivers to get out of my way"

Nate S
2nd December, 2011 @ 12:40 pm PST

How about just looking both ways before flooring the gas pedal for good measure??... and denying a driver licence to people unable of doing such a simple task....

Bernardo Espinosa Larracoechea
2nd December, 2011 @ 05:26 pm PST

The driver who runs red lights is a danger not only to himself, but particularly for other road users such as pedestrians and cyclists. A driver who does so consciously takes a risk carrying its heavy consequences.

Such a warning system would engage its user to pay more attention to potential dangers from drivers who regularly terrorize the roads allowing him to protect himself and the passengers.

Fretting Freddy the Ferret pressing the Fret
13th November, 2012 @ 11:15 am PST
Post a Comment

Login with your gizmag account:

Or Login with Facebook:


Related Articles
Looking for something? Search our 29,015 articles