One of the worst things about traffic is that it’s so unpredictable. You can be whizzing along one minute, and crawling with the snails the next. Even the real-time traffic information that a few companies, notably Google, now provide, can be obsolete by the time you’re on your way.But a small cadre of lucky San Franciscans will soon be finding out where the traffic will be before it happens, thanks to a joint project by the California Center for Innovative Transportation (CCIT) at the University of California, Berkeley (my alma mater) the California Department of Transportation, aka Caltrans, and IBM.
The principle behind the system is that when accidents happen, the traffic jams up in a predictable manner, “like a shock wave in a fluid,” Alexandre Bayen of CCIT tells New Scientist Magazine. By combining these patterns with real-time traffic data, the system is able to predict congestion up to 40 minutes before it even happens.
The hardware for this initiative consists of “inductive loop sensors” in the roadways. The data so-gathered is fed into IBM’s Traffic Prediction Tool. Data from users’ GPS units and smartphones on their usual routes and travel times completes the circle, enabling the system to provide alternative routes. And it’s getting more sophisticated all the time, learning to relieve congestion without creating new chokepoints, a skill requiring the ability to predict the effects of congestion-relieving redirections.