MIT Algorithm Predicts Red Light Runners 348
adeelarshad82 writes "Researchers at MIT have developed an algorithm that determines which drivers will run a red light, within one to two seconds before a potential collision. The research, based on 15,000 cars at a busy intersection, monitored various factors to determine which cars were were likely to run a red light. They found that their predictions were correct about 85 percent of the time, which is about 15-20 percent better than existing traffic prediction algorithms."
Re:Just a matter of time... (Score:4, Informative)
For anyone confused; a bollard is a retractable concrete or metal post that comes out the ground to block traffic. They seem to be popular in Europe.
http://www.youtube.com/watch?v=KIqlkPhDfwM [youtube.com]
http://www.youtube.com/watch?v=4ZdLjKl0lHc [youtube.com]
Re:Just a matter of time... (Score:4, Informative)
Who cares if we can "catch" more people?
The people who add the fines to their revenue.
As far as I'm aware, the only thing that's been proven to reduce the number of accidents at stop lights is to make the orange phase longer. This is why cities that want to increase revenue have often been found to have made the orange phase shorter.
Re:Just a matter of time... (Score:5, Informative)
Of course, you are making an error of assumption in assuming that people who run lights generally do it willfully by thought, and not negligently by distraction, or though misjudgment.
Actually, thats one of the few things that I remember from taking the one social psych course that I took.... they called it the "fundamental error of assumption". That is, that people tend to ascribe internal motivations to other people's actions, and external ones to our own. So, you ran the red light because you are impatient and try to cut it as close as you can. I ran the red light because the yellow was excessively short, and you were sitting in the passenger seat talking to me and distracting me.
Sounds ridiculous when you say it like that but, its actually pretty common.
Re:Where's the Work? (Score:4, Informative)
The paper is here [mit.edu], and it gives ROC curves. They used two approaches, a hidden Markov model and a support vector machine Bayesian filter.
Re:Just a matter of time... (Score:5, Informative)
"they called it the "fundamental error of assumption""...
I think you mean the fundamental attribution error [wikipedia.org]?