Using an internet connection, the car talks to city infrastructure services to figure out the timing of upcoming traffic lights. It then uses GPS and speed data to work out exactly what speed you need to drive at to catch an upcoming green light, and shows it on your speedo as a nifty little green zone you can aim for.
Obviously, it can't take into account any traffic that might have backed up. And it won't save you from every red, unless you're prepared to trickle along at a walking pace and endure a symphony of beeping from the drivers behind you.
But it's a nice interface for a system that could save drivers a bit of fuel and a bit of frustration as we move toward a world where all this stuff happens automatically in autonomous cars.
Other initiatives under testing alongside GLOSA include real-time parking spot tracking, emergency vehicle alerts, and an interesting intersection management system that informs drivers if other cars are coming from either side and can "suggest the order in which cars should proceed at a junction" – clearly something that'll offer huge opportunities in any all-autonomous zone.
Indeed, one wonders how the role of traffic lights will change in an entirely self-driving environment where cars should theoretically be able to proceed full speed through intersections, all cars communicating to pick the gaps and avoid collisions. Perhaps the only function of traffic lights might be to help out pedestrians and bicycles. But in the meantime, a system like GLOSA could help drivers stay in the zone to avoid those annoying red lights.