The research is being published in the journal IEEE Transactions on Intelligent Transportation Systems. The team measured the success of their algorithm with 15,000 vehicles at an intersection in Christianburg, VA, and the group found that the algorithm was correct 85 percent of the time, which is reportedly the most accurate that has been measured.
While this information doesn't do much now for the safety of others, the smart cars of tomorrow could do a lot with that info. Since the information can be calculated in milliseconds, if the information can be transmitted to other vehicles, those vehicles can take decisive measures on the behalf of the driver. MIT Professor Jonathan How explains that, even though a driver sees a green light, a heads-up display could warn the driver of an impending red light runner.