Comment: Re:Already implemented here (Score 2) 297
Oh, they do something... just not at the time of day that you're driving. Try driving at three or four in the morning some time, and you'll see just how well those sensors work.
What this professor proposes is basically a massively scaled-down version of what I've been proposing for years. Unfortunately, that scaled-down nature of the proposal makes it a lot less useful in practice. To do traffic optimization well, you really need automated vehicles so that people register their destinations with a central system that can optimize which roads each vehicle takes, optimize which lanes go in which direction, and optimize when vehicles should pass through intersections to minimize stopping. The more information you have, the easier it is to make such decisions. More to the point, by knowing the entire route (rather than just one or two intersections ahead), you can do a much better job of optimization.
For example, if you know that a vehicle passes through three traffic lights in a short period of time, you may find that by making the vehicle stop at the first light rather than the second light (or vice-versa), you shift its arrival at the third light enough so that a vehicle does not have to stop that otherwise would have stopped, resulting in an overall efficiency gain.
Eventually, when nearly all cars have been converted to automatic drivers, you could leave the traffic lights in place, with all directions red by default, turning green only when a legacy manual vehicle approaches. Until then, however, having ten or twenty seconds of advance warning won't really help all that much. As others have said, we already have road sensors for this. If the lights are configured to not use the road loops, the operators are sure as heck not going to upgrade the lights to use a transponder-based system that gives them even more inputs to ignore.