Researchers at MIT and Princeton have developed the Signal Guru app, which you can read about here. Us traffic engineers are used to working on the supply side of the transportation capacity equation - adding more lanes or tweaking signal timings to improve traffic flow. The Signal Guru allows drivers to work on the demand side of the equation.
Basically, if a lot of people are using the Signal Guru app, the computer that's receiving all of their GPS data can figure out the signal timing in the area. This data is then fed back to the driver's smartphone to tell them if they should slow down or speed up to better sync with the signal timing plans. Based on the initial trials done by the researchers, the drivers cut fuel consumption by 20% by driving per the Signal Guru directions.
Our traffic signals (at least in Minnesota) are very "wired." Plus agencies are collecting a tremendous amount of traffic flow data with different types of traffic sensors. You can pull up traffic flow maps with any smart phone to see current traffic conditions (the roads are typicaly coded green, yellow, or red - good to bad).
If, and it's a big if, the researchers could tap this data stream to power the Signal Guru instead of relying solely on other Signal Guru users, a much more robust system could be built. Anyone at Google listening?