Computer to predict traffic jams

Traffic websites, with their color-coded maps of clogged streets and freeways, are good at telling commuters when congestion is already awful. But what if they could know not only where you drive, but if the route is going to be bad today, and warned you ahead of time?

A team of researchers at IBM Corp. are working on a system that would do just that.
They've combined sophisticated analytics software with a network of sensors the state has already embedded in roadways throughout California. With the help of a database of past traffic tie-ups, they say they can predict when they'll happen in the future.

IBM, which worked with the state Department of Transportation and the California Center for Innovative Transportation at the University of California, Berkeley, can't tell you when an accident is going to happen -- yet. But John Day, an IBM manager, says that's a natural extension of the work his team has been doing, and one day could well be something they could reasonably infer from the data they're collecting.

For now, the researchers are showing that they can go a step beyond what's available on regional traffic websites, which report current conditions. The most that might tell you is that your drive is already backed up.

"Instead of just showing them look how ugly this looks, or what it looked like five minutes ago, let's give them an idea of what this looks like 30 to 40 minutes from now," Day said in an interview.

IBM's program relies on a tactic some might find invasive -- using GPS coordinates from users' cellphones -- to learn their daily commutes so it can offer suggestions when those routes are clogged. IBM emphasizes that users would need to give permission to the tracking. (Google Inc.'s traffic-monitoring program uses locations reported by cellphone chips, but only to figure out if there's congestion.)

Day says the project, which is still in the research stage, could eventually be combined with other data such as train schedules and other transit data, to provide a more complete picture of people's daily migrations and take a constant pulse of the health of a region's transportation infrastructure.