Called SignalGuru, the mobile app uses the cameras on cellphones to crowdsource traffic signal data and relay to users exactly how fast they need to drive to avoid red lights. Developed by researchers at the Massachusetts Institute of Technology, the app analyzes the data and, using the GPS in the smartphone, calculates the distance to the signal and optimal speed.
The first prototype was released in Nov. 2010, and the project involved collaboration between MIT, Princeton University and the Singapore Future Urban Mobility Center.
Emmanouil Koukoumidis, a computer scientist and the app’s lead developer, said the initial goal was to help make urban transportation more efficient. He explained that traffic signals force drivers to come to a complete stop, and as a result of the acceleration needed when a signal turns green, fuel gets wasted, traffic flow is interrupted and emissions increase from vehicles.
So the research team focused its effort on using technology to allow people to regulate their travel speed to get through intersections unimpeded. Koukoumidis said that proposing new hardware for inside a car could take “forever” for the audio industry to integrate, so he concentrated on technology most people already have — the sensors embedded in mobile phones.
For the app to work, the mobile device has to be mounted on a vehicle’s dashboard — or on the windshield — and the camera will capture all the data needed to estimate the times. Tested on standard traffic signals in Cambridge, Mass., the app was accurate within .66 of a second. When tested against more advanced adaptive traffic control signals that adjust the length of signals according to the flow of traffic, the app was accurate within two seconds.
Koukoumidis explained the biggest novelty of the system was crowdsourcing information from the mobile device cameras.
“There have been research efforts that were applications that used GPS or the accelerometer of the phones to crowdsource and build an interesting service,” Koukoumidis said. “But this was the first system that said we can use the cameras to crowdsource and build a service on top of that.”
Further Work Needed
While the app has promise, Koukoumidis acknowledged a few things need to happen before it’s ready for public use. First and foremost, the research team needs an industrial partner that can commercialize the idea.
From a usability perspective, the way the system displays the speed a driver must travel to make a green light also needs to be improved. Right now, it’s just visual. In order to make things practical, an audio advisory must be built into the app so that a driver doesn’t have to look at their phone while driving.
Another concern is users driving faster and exceeding the speed limit to get through an intersection before the light changes. Koukoumidis said his idea is that the system can be tweaked to shut down if it notices someone driving over the speed limit regularly.
“Because the system knows which intersection you are approaching and when the light will change, it can easily tell that the person is driving over the speed limit to beat the light,” Koukoumidis explained. “If it sees you are using the advisory in a way it isn’t intended to, it can disable [it.]”
Photo courtesy of Emmanouil Koukoumidis
