ROBERT SIEGEL, HOST:
From NPR News, this is ALL THINGS CONSIDERED. I'm Robert Siegel.
Car companies have already begun to design cars that can drive themselves. But to make these smart cars really useful, they'll also need smart roads. As part of his series, "Joe's Big Idea," NPR science correspondent Joe Palca has this story about some computer scientists who were designing a smart traffic intersection. How smart? Well, it can keep traffic flowing at least 10 times faster than old-fashioned intersections.
JOE PALCA, BYLINE: I've done my share of late night driving, so I can sympathize with what happened to Kurt Dresner one night about a decade ago. At the time, Dresner was a computer science grad student at the University of Texas at Austin. He was working in the lab late one night - did I just make you think of a song? - and didn't head home till nearly 2 in the morning.
KURT DRESNER: I was driving home, and I was at a stoplight. And the stoplight was red but there was nobody around. And I was just thinking to myself, there's got to be a better way to do this.
PALCA: The better way Dresner came up with requires autonomous cars - cars that know exactly where they are, precisely how fast they're going, and can drive themselves without any help from a human. It also requires smart intersections. In his system, Dresner says, as a car approaches the intersection, it would request a reservation to enter the intersection by sending ahead detailed information.
DRESNER: This is how big my vehicle is, this is when I'm going to enter the intersection, this is how fast I'm going to be going, this is how I am going to exit the intersection and when I'm going to exit the intersection.
PALCA: A computer at the smart intersection looks at this information and information it's getting from other cars wanting to enter the intersection, then makes a decision.
DRESNER: All the intersection ever says is yes or no. Either it says, yes, that works and I've confirmed your reservation or it says, no, that doesn't work and you need to come up with something else.
PALCA: So the car can propose speeding up or slowing down or stopping for a while until it finds a plan the intersection will accept.
DRESNER: It's up to the car to change the plans.
PALCA: I've seen a simulation of how this would work. On the screen, there's a four-way intersection with each road carrying five lanes of traffic. With light traffic, cars just zoom through. But with heavier traffic, even though some cars are turning and some are going straight, in the simulation at least, the cars don't slow down all that much, although they frequently come terrifyingly close to each. And this is why you need autonomous cars. Dresner says a human driver just couldn't make the necessary adjustments fast enough or remain on course reliably enough to make the system work.
DRESNER: You have to trust the vehicle at that point, and the vehicle can do it a lot better than you can. I mean, I don't - I can't count the number of times I'm driving around and I think to myself, this would be so much better if there were a computer driving these cars around me because the people around me are not driving very well.
PALCA: Dresner knows right now most people wouldn't be comfortable putting themselves or their families into the hands of a car that drives itself.
DRESNER: My personal feeling is when you're neighbor buys that autonomous vehicle and you're seeing them not have to worry about driving when they're tired or getting a designated driver for a party or having the car drive their kid to soccer practice without them having to wake up, you know, early on Saturday morning or whatever, personally, I think you'll be sold.
PALCA: He may be right, but we're probably not going to find out for a while. Joe Palca, NPR News.
NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.