How Uber is helping steer the future of self-driving cars
GWEN IFILL: Now: The Jetsons future may be arriving sooner than you think, for better or for worse.
Uber is experimenting with self-driving cars. In Pittsburgh today, the company began deploying a small test fleet of the vehicles around the city.
Hari Sreenivasan has the story.
HARI SREENIVASAN: OK, the first thing to know, Uber tried this with several journalists this week. Each self-driving car was accompanied by a human operator, who loosely kept hands on the steering wheel.
The cars are equipped with sensors, radars and light-mapping systems. Select Uber customers in Pittsburgh will be able to opt into a driverless car pickup.
Alex Davies writes about all things transportation for "Wired" magazine, and took a ride in one of Uber's self-driving cars. He joins me now from San Francisco.
So, Alex, unless you were a friend of a Tesla driver or a Google engineer, you're one of the first people to sit in the back of one of these cars. What was the experience like?
ALEX DAVIES, Wired: So, for the most part, it was actually kind of just like a regular Uber ride, minus the fact that it was clearly a kind of carefully orchestrated media preview.
It's same way most Uber rides start. You pull out your phone, you open up the Uber app, and you hit — enter your destination. Then you call up the car. And then what's going to happen from now on for some select customers in Pittsburgh is, it will say, hey, would you like a self-driving car, instead of some guy driving in the street who's trying to make extra money?
And if you hit yes, then that's what shows up. And basically a car — right now, they're now using Ford Fusions covered in this enormous pile of sensors on the roof, spinning light scanners, radars, cameras. That pulls up.
But after that, once you get in, it works more or less like a normal Uber. You relax in the back seat and the car drives you exactly where you're going.
HARI SREENIVASAN: Now, we mentioned the biggest safety precaution, which has been a human, an engineer, a driver in the car. But what other kinds of precautions has Uber taken to roll this test out?
ALEX DAVIES: So, first of all, they're only operating within a small select area of downtown Pittsburgh. They're open to about 12 square miles right now.
And the reason they're limited to that space is that they will only send their cars out to areas that they have mapped in extreme detail. That means that the car already knows exactly where every traffic light is. It knows which lanes are right-turn-only lanes, which — where it's allowed to make a U-turn or where it can't, what the speed limit is everywhere, and where pedestrians are likely to cross, where cars are usually driving if the lane lanes aren't super well-marked.
So, basically, it's kind of slicing off the riskier areas by only operating in places where it knows what the conditions are going to be like.
HARI SREENIVASAN: So, why Pittsburgh?
ALEX DAVIES: A couple of reasons.
The biggest one is that Pittsburgh is home to Carnegie Mellon, which has — Carnegie Mellon has one of the best robotics programs in the country and probably in the world. And some of their engineers have been studying self-driving cars for 15 years now, well before anybody imagined that this could really be a thing.
So, a lot of those guys and women are now working for Uber at its Advanced Technology Center in the city. The second reason is that Uber has — sorry — Pittsburgh has a lot of different weather conditions, unlike Silicon Valley, which is pretty much always sunny.
Pittsburgh, you're going to have to face snow and rain and different weather problems. And that's a good challenge for the cars to learn how to solve. And the street grid isn't the easiest to navigate. So, again, it's training cars to take on more complicated situations.
The third advantage to Pittsburgh is that Pennsylvania hasn't really regulated this space yet, so Uber is free to do pretty much what it wants.
HARI SREENIVASAN: What is the safety record of autonomous vehicles to date, and how is Uber factoring that in to these tests?
ALEX DAVIES: So, overall, the safety record is excellent, because, for the most part, at least when you're talking about fully autonomous vehicles, they have always got a man — a trained operator at the wheel ready to take over in any situation.
So, for example, Google, which tests constantly in Mountain View and in Austin and a few other places, has had a few minor accidents, but nothing serious, nothing where anyone's ever been injured.
And the same thing with Uber, at least in California, where it's required to report any accidents, as are all autonomous car operators. In Pennsylvania, it's not exactly clear what the record is, though they say they haven't had any accidents, just because there are no rules saying they have to tell you if there is one.
HARI SREENIVASAN: All right.
And, also, put this into perspective. Why is Uber doing this? We have heard about Google. We have heard about Tesla. We have heard about auto manufacturers. What's Uber's interest in having autonomous vehicles?
ALEX DAVIES: So, Uber's interest is natural, because when you ask automakers, for example, well, when you think about an autonomous fleet of cars, what does that look like, they will tell you, well, it looks like Uber. It's a car that you don't have to own or deal with parking or maintenance that shows up, picks you up and takes you where you want to go.
So if Uber can take its model, which is already enormously successful and rapidly spreading around the world, and it can take out the single most expensive part of that, which is a human driver who takes half or three-quarters of every customer's fare, and they can remove that person from the equation, then, all of a sudden, the business gets a lot more efficient and a lot more profitable.
HARI SREENIVASAN: OK. And how far out is that future?
ALEX DAVIES: So, Uber right now, if you say, well, when are you going to be able to take the engineers out of the cars, they will say, when it's safe, which is not a particularly helpful answer.
But if you look at the timelines other players have put out, Baidu, which is kind of the Chinese equivalent of Google, says it wants self-driving cars on streets in 2019. Ford is targeting 2021.
So, I think it's safe to say that Uber will be more or less on that timeline, so three to five years out.
HARI SREENIVASAN: All right, Alex Davies from "Wired" joining us from San Francisco tonight, thanks so much.
ALEX DAVIES: Thank you.