Algorithm Predicts Auto Traffic Intersection Violators

(A. Kotok)

Aerospace engineers at MIT have developed an algorithm tested on real-life traffic data that predicts when an oncoming car is likely to run a red light at an intersection. The team’s research is expected to appear in an upcoming issue of the journal IEEE Transactions on Intelligent Transportation Systems.

Researchers from MIT’s Aerospace Controls Laboratory, led by aeronautics professor Jonathan How, use data on a vehicle’s deceleration and its distance from a light to determine which cars were likely to cross into an intersection after a light has turned red, and those cars that would obey the signal. The ability to judge the intentions of other drivers has real-life implications. In 2008, according to the National Highway Traffic Safety Administration, 2.3 million automobile crashes occurred at U.S. intersections, resulting in some 7,000 deaths, with more than 700 of those fatalities were due to drivers running red lights.

The MIT engineers designed the algorithm for cars of the future with driver assistance systems that can warn drivers of a potential collision based on another driver’s behavior. How says that to implement this kind of warning system, vehicles would need to be able to wirelessly send and receive information such as a car’s speed and position. This vehicle-to-vehicle communication, he says, can potentially improve safety and avoid traffic congestion. The U.S. Department of Transportation (DoT) and Ford Motor Company are exploring prototypes based on Wi-Fi and collision-avoidance systems.

Georges Aoude, a former graduate student of How’s, designed the algorithm based on a technique from artificial intelligence domains that captures a vehicle’s motion in multiple dimensions with a highly accurate and efficient classifier that can be executed in less than five milliseconds. The researchers tested the algorithm on data collected from a busy intersection in Christiansburg, Virginia, which the DoT outfitted from another project with instruments that tracked vehicle speed and location, as well as when lights turned red.

Aoude and colleagues from MIT applied their algorithm to data from more than 15,000 approaching vehicles at the intersection, and found that it was able to correctly identify red-light violators 85 percent of the time, an improvement of 15 to 20 percent over existing algorithms. The algorithm also accurately identified potential violators within a couple of seconds of reaching a red light, enough time, according to the researchers, for other drivers at an intersection to be able to react to the threat if alerted.

Compared to other efforts to model driving behavior, say the researchers, the MIT algorithm generated fewer false alarms, an important advantage for systems providing guidance to human drivers. Other algorithms, says How, tend to be overly cautious, which can lead to drivers ignoring their advice.

A video from the Aerospace Controls Lab shows the algorithm in action on a model intersection.

Read more:

• Simulation Improves Safety at Traffic Intersections
• Engineers Developing Human Driving Model to Reduce Crashes

*     *     *