Predicting Driver Behaviour – Drivers who Run Red Lights

We all know that not everyone who signs up for a car finance package will be a responsible driver.  It is hard to distinguish the good ones from the bad ones before they get on the road – until now, that is.  MIT researchers believe they have successfully developed an algorithm that can predict driver behaviour at intersections with a high degree of accuracy – including whether that driver is likely to stop at the red light or not.

According to the two classes of algorithms drivers are classified as “violating” or “compliant.”  The algorithms are based on hidden Markov models and support vector machines that have been used in multiple disciplines with a remarkable degree of accuracy.

Figures supplied by the National Highway Traffic Safety Administration show that 37,261 die in road accidents in the United States every year.  For the UK this figure is between 2,000 and 3,000.  Forty-five percent of injuries and 22% of fatalities take place at intersections.  Even more worrying: 50% of the people killed in these accidents are not those who run red lights, but pedestrians and the occupants of other vehicles

The results of the extensive study of driver behaviour will soon be published in a paper in the journal IEEE Transactions on Intelligent Transportation systems.  According to the findings it is possible to use statistical models to predict how a driver will react when he or she is faced by a red light.  Factors taken into account include driver perception, reaction time, age, distance to the intersection and speed of the vehicle.

Got a question for us?

If you require further assistance, our team of Customer Advisors are here to help. We're open six days a week - you can view our opening hours here - and we're more than happy to answer your questions.