美国国家公共电台 NPR Self-Driving Cars Could Ease Our Commutes, But That'll Take A While（在线收听）

DAVID GREENE, HOST:

All right, here's something that will come as no surprise to anyone who drives to work. Maybe you're driving to work right now. Commuting times are just getting longer and longer. The Census Bureau says the average commute takes 26 minutes. That is 20 percent longer than in 1980. NPR's Yuki Noguchi is asking if self-driving cars, which might improve traffic flow, could shorten her commute.

YUKI NOGUCHI, BYLINE: I'm standing here watching rush hour build. I'm overlooking the place where Interstate 66 and the Dulles Toll Road merge just outside Washington, D.C. It's here I meet Hesham Rakha, an engineering professor at Virginia Tech with a lifelong interest in traffic - how it flows and why it sometimes doesn't.

HESHAM RAKHA: Yeah, that's 66. It's 66.

NOGUCHI: It is, he says, extremely easy for human drivers to create traffic jams.

RAKHA: One little touching of the brake can pull the whole system down.

NOGUCHI: A driver taps his brake, subsequent drivers react, each braking harder than the next. Soon, traffic is crawling for no apparent reason.

RAKHA: So if we can prevent that congestion from happening, not only will we let vehicles travel faster, but we'll also get more vehicles through our system.

NOGUCHI: Self-driving cars can synchronize. They communicate with one another, signaling disturbances ahead, avoiding backups. To illustrate how autonomous cars flow differently, Rakha pulls up a traffic simulator on his laptop.

RAKHA: This is what we call a fundamental diagram.

NOGUCHI: The simulator uses recent actual traffic data collected from the highway merge below us.

RAKHA: And the red is the congestion which spills back from this.

NOGUCHI: An algorithm then allows Rakha to simulate how the colored dots flow when some or all of the cars are autonomous. The biggest bugaboo of traffic - accidents caused by human error - go down dramatically. On the screen and we can see how the red dots indicating congestion gradually clear and become green moving dots.

Geoff Wardle teaches transportation design at the ArtCenter College of Design. He says he expects more people will share automated cars, creating less demand for parking, or the cars might park themselves. Either way...

GEOFF WARDLE: A significant amount of traffic congestion is caused by people going round and round the same two or three blocks looking for that place to park.

NOGUCHI: That's the positive side of self-driving cars. Here's the negative. Again, here's Wardle.

WARDLE: Human beings, we tend to like to try and cheat the system to gain personal advantage.

NOGUCHI: In fact, Virginia Tech engineer Rakha's research shows that once self-driving cars make their way onto our roads, it will take a while to see any improvements in traffic. The behavior of self-driving cars is predictable, but human behavior is not. So it's not until there are a lot of driverless cars on the road that we might see some easing of congestion. Put another way, some humans are jerks who try to cut in at the front of the line. Oh, you know who I'm talking about - Rakha certainly does.

RAKHA: And they'll cut across all the lanes and cause everyone delay just because they want to save a couple of seconds. They break down the whole system.

NOGUCHI: When people ask you so are automated cars going to reduce my commute time, what do you tell them?

RAKHA: I don't know the answer. If the road is less congested, more people are going to be attracted to that road and so basically will become congested because it's supply and demand.

NOGUCHI: If traffic does improve, in other words, more cars and trucks are likely to use the road. And for us right now...

Maybe we should get out of here before traffic gets bad.

RAKHA: Yeah, I think it's a good idea.

NOGUCHI: Yuki Noguchi, NPR News, Washington.

(SOUNDBITE OF REKI SONG, "THE ORIGIN")