Slashdot Mirror
NASA's Sensor Web
ddtstudio writes "PC Mag has a story about the Sensor Web: 'a cutting-edge application of networked sensor technology currently on the fast track at NASA's Jet Propulsion Laboratory (JPL).' Not only a new way to test tech, but also perhaps a pervasive and inexpensive way to explore remote places such as Antarctica -- or Mars."
IANATE (I am not a Traffic Engineer), but I am the son of one. As somebody already replied to you, they have loops under the asphalt at most intersections. They're basically big metal detectors (which always made me want to try to degaus someone's car so they wouldn't trip the signals). They're hooked into signal boxes on the corners and sometimes into central monitoring and control centers. They are used primarily to control signals rather than modeling traffic flow for study purposes (for that they mostly use a pneumatic tube running across the road which trips a counter). Signals can be operated by either loops, timers, or a combination of the two -- often times that varies by day, for instance a light at a crossroad may be timed during the day, but, at night, may be set to change only if someone pulls up to the loop. This is one reason why it pisses me off when someone stops ten feet before the stopbar (just before which the loops are generally located) -- that and it just seems like a stupid-ass thing to do.
As far as optimal stoplight pattern, that's a tricky issue. In many cases, lights are timed or operated by loops such that overall average vehicle travel time is greater than it could be, simply because studies have shown that driver satisfaction is increased if travel on major roads is less impeded by signals, even at the expense of greater delays on side roads. It's a matter of perception. So the 'optimal stoplight pattern' is a combination of decreasing average vehicle travel time and accomdating drivers' faulty perceptions of what is optimal.
As is often the case with human interaction with technology, the technology here is well-studied and well-tuned. The problems mostly lie with human perceptions, reaction time, errors, etc. For instance, the average time between when a signal turns green and when the first driver begins to move is 3 seconds. That also pisses me off. I drive a stick and I'm gone in a second.
Also, having said all that about the technology being up to speed, I have to qualify it. The states set up regulations on signals, including their timing and such, but it is up to the government which owns the signal to implement the policies. In many cities there is no central signal administration and signals operate independantly or in small networks which means poorer coordination of traffic. Also, city traffic engineers are not always Traffic Engineers but often electrical or other engineers. No doubt they are fine electrical engineers, but you wouldn't necessarily want one designing bridges, nor managing traffic.
So the answer to your question is that the problem is not with the modeling or the engineering, but rather with the implementation.