The continuing increase in gasoline prices around the world over the past decade has also seen an increase in the practice of hypermiling - the act of driving using techniques that maximize fuel economy. One of the most effective hypermiling techniques is maintaining a steady speed while driving instead of constantly stopping and starting. Unfortunately, traffic lights all too often conspire to foil attempts at keeping the vehicle rolling. Researchers at MIT and Princeton have now devised a system that gathers visual data from the cameras of a network of dashboard-mounted smartphones and tells drivers the optimal speed to drive at to avoid waiting at the next set of lights.

The new system, dubbed SignalGuru, was tested in both Cambridge, Massachusetts, and in Singapore. In Cambridge, where traffic signals are on fixed schedules, the researchers say the system was able to predict when lights would change with an average error of only two-thirds of a second and helped drivers cut fuel consumption by an average of 20 percent. In Singapore, where the duration of lights varies continuously according to changes in traffic flow, the error increased to an average of slightly more than one second, with one particularly light in densely populated central Singapore seeing an average error of more than two seconds.

The version of the system used in the tests graphically displayed the optimal speed for avoiding a full stop at the next light, but a commercial version would probably use audio prompts said Emmanouil Koukoumidis, a visiting researcher at MIT who led the project. The researchers also modeled the effect of instructing drivers to accelerate in order to catch lights before they changed, but decided that wasn't the safest option.

"The good news for the U.S. is that most signals in the U.S. are dummy signals," (signals with fixed schedules), says Koukoumidis, who launched the SignalGuru project at MIT with Li-Shiuan Peh, an associate professor in the Department of Electrical Engineering and Computer Science who came to MIT from Princeton in fall 2009. But Koukoumidis says even an accuracy of two and half seconds, "could very well help you avoid stopping at an intersection." He also points out that the predictions for variable signals would improve as more cars were outfitted with the system, collecting more data.

Koukoumidis says cars are responsible for 28 percent of the energy consumption and 32 percent of the carbon dioxide emissions in the U.S. and that, "if you can save even a small percentage of that, then you can have a large effect on the energy that the U.S. consumes."

But it isn't just more economical driving that could benefit from the technology. Koukoumidis says the computing infrastructure that underlies the system could be adapted to a variety of applications that could be useful to commuters, such as capturing information about prices at different gas stations, the locations and rates of progress of city buses, or about the availability of parking spaces in urban areas. The system could also be used in conjunction with existing routing software, to recommend ducking down a side street instead of slowing to a crawl to avoid a red light, for instance.

The researchers from MIT and Princeton took out the best-paper award for the SignalGuru system in July at the Association for Computing Machinery's MobiSys conference.