SignalGuru app helps drivers avoid red lights

TORONTO Mon Sep 17, 2012 5:44am EDT

Related Topics

TORONTO (Reuters) - Researchers at the Massachusetts Institute of Technology have developed a system that uses dashboard-mounted smartphones to help drivers avoid red lights and reduce fuel consumption.

The app called SignalGuru predicts when a traffic signal is about to change, and the speed that should be driven when approaching an intersection in order to cruise through without stopping.

"The stop-and-go pattern that traffic signals create increases fuel consumption significantly," said Emmanouil Koukoumidis, the scientist behind the app.

"We wondered how we could help drivers cruise through signal light intersections without stopping, and how much we could save on gas and improve the flow of vehicles," he added.

When approaching an intersection, the camera on a driver's dashboard-mounted smart phone is activated, which detects when a signal transitions from red to green and vice versa.

Using this information, the app determines the speed that should be driven to avoid stopping at a red light on the cusp of turning green, or a green light just shy of turning red.

"It tells the drivers that 'if you drive at 30 miles per hour then you'll be able to cruise through without stopping,'" explained Koukoumidis, adding that the speed recommended is always within legal speed limits.

Information on the traffic signals, such as when they change, is crowdsourced by other users of the app and then sent back to SignalGuru to improve the accuracy of its predictions.

Koukoumidis said that while testing their prototype in Cambridge, Massachusetts they saw a 20 percent decrease in fuel consumption, which could have a significant monetary and environmental impact.

"In the U.S. we're spending 1/3 of the annual energy consumption for transportation and a big part of that is vehicles," he explained.

The system was also tested in Singapore, where the traffic lights vary depending on the volume of traffic.

"It was less accurate compared to Cambridge where signals were pre-timed and had fixed settings but it would still work reasonably well with predictions accurate within two seconds," Koukoumidis said.

Crowdsourcing information about signal lights is necessary, he said, because this data is difficult to access from traffic authorities, which are not unified and do not always have the information computerized.

But this could also pose safety concerns, for example, a signal not changing when predicted due to inaccuracies.

"SignalGuru will advise the driver when to arrive at the intersection but the driver should always check for himself that the light indeed turned green," he said, noting that it's similar to how a driver does not follow a navigation device blindly.

Currently the group is looking for industrial partners to commercialize the software. They also plan to implement other safety features, such as thresholds on deceleration, before making it accessible to the public.

Koukoumidis said that going forward their patented approach could also be used to capture other information about the real world, such as available parking spaces or real-time gas prices.

" are computer eyes looking out into the street that can capture all sorts of information," he said.

The research project was launched as part of the Singapore-MIT Alliance for Research and Technology's Future Urban Mobility group, in which professors Margaret Mantonosi and Li-Shiuan Peh were advisors.

(Editing by Paul Casciato)

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see
Comments (2)
Numb3rTech wrote:
Very good idea and about time something like this was promoted and built into cars. Should have been done a few years ago. Would really save fuel if that happened. Good going to the people doing this.

Sep 17, 2012 5:32pm EDT  --  Report as abuse
jhvance wrote:
This app will be effective primarily in US cities whose signal system timing patterns are non-adaptive, unlike most European, Asian and Australian cities which use newer adaptive signal system technologies. Some US cities are moving to implement adaptive technology, but their legacy systems long in place were (and still are because of budget limitations) built around central computer integration with the local controllers based upon fixed timing plans for progressive flow along dominant arterial streets derived from traffic count patterns during morning peak, midday off-peak and evening peak periods (and perhaps night/weekend periods). However, in many US cities these timing plans have become outdated due to reduced local government staffing and budgets for gathering updated traffic counts periodically and a general transition toward outsourcing signal hardware maintenance contracts which don’t incorporate requirements for updating the integrated timing characteristics.

While this current app might have some near-term utility in the US, as the urban areas replace their older signal system technology with the newer adaptive variety and the convergence of enhanced communication between vehicles and traffic control systems becomes more deeply embedded with both public- and private-sector R&D efforts in support of broadly-distributed capability for real-time exchange of basic data about the traffic system state, it’s very likely a future version of the app running on an even more enhanced smartphone (or directly within the vehicle’s control and communications center) will utilize those distributed external sources of information rather than rely upon the sole-perspective visual reading. Ultimately, an increasing proportion of the vehicle fleet that becomes capable of such enhanced communications and autonomous operation could produce an integrated system approaching the “seamless” ideal that app strives to establish.

Sep 17, 2012 5:59pm EDT  --  Report as abuse
This discussion is now closed. We welcome comments on our articles for a limited period after their publication.