Aurora Is Not Building Autonomous Cars, It’s Building Safe Drivers

Aurora Is Not Building Autonomous Cars, It's Building Safe Drivers

Chris Urmson, CEO of Aurora Chris Urmson was designing self-driving vehicles at Carnegie Mellon years before Uber, Lyft, or Waymo were founded, let alone hit the open roads. He was part of the team that won the 2007 DARPA Grand Challenge and served at the CTO of Google’s self-driving car program.

Today, he is CEO of Aurora . You won’t see an Aurora-branded car on the road any time soon, but the company is quickly gaining a reputation as the firm that will make autonomous cars actually work. Urmson was at SXSW in Austin to help promote Autonomy, a self-driving car documentary in which he appears, and that’s where we caught up with him.

Dan Costa: I was watching the Autonomy documentary last night and you show up in the first five minutes talking about self-driving cars. You say that one of the reasons you’re in this business is so your son will not have to get a driver’s license. Can you just talk a little bit about how that motivates you and what Aurora does?

Chris Urmson : Aurora [is] building self-driving car technology. We don’t build the car. We don’t think of really about building the application, or ride-hailing or whatnot. We think about how do we build a really safe driver. So we’ve been at it for a couple years, we’re a couple hundred people at this point.

And really, what gets us up in the morning is all the benefits you can see from this technology. We can save lives on the road, we can make transportation more accessible, we can make cities more livable. For me, I think I have two awesome sons. And if you look at the kind of fatality curves for driving, when it comes to age and the probability that something terrible happens, it looks like a bathtub. The youngest new drivers, and then old drivers, they’re in these accidents more often. So getting this technology out in the world so that young people like my kids don’t have that risk, parents don’t have to worry about it, that is exciting and meaningful. Dan Costa: I think that’s an important point, that there’s two sort of reasons for building self-driving cars. The engineering drive just because we can. We have a problem we know that we can solve it and we can solve it with engineering, but fundamentally this is a safety issue. And there’s 40,000 deaths every year and the vast majority of those is because of human error. This is what this technology is designed to prevent.

Chris Urmson : That’s exactly right. So in America alone, 40,000 people every year, 1.3 million globally. That’s incredible. Something like 2.5 people a minute die in traffic accidents around the world. Ninety-six percent of those accidents are due to human error. So that means that we can do something about it, right? We can build technology that is always paying attention to the road, that isn’t kind of figuring out whether there’s a new text message that came in. Or being distracted in the car, or just falling asleep, or had too many drinks. It’s technology that is paying attention the whole time and just as good the whole time when it’s operating. And I think that’s incredible.

I feel very lucky to work in a space, like you said, that the technology itself is just cool. It’s broad and interesting and it’s a neat problem. It’s tangible, right? You can go touch the car, you can see it when it gets better. But then it has this opportunity to have a profound impact. Again, in safety, but transportation touches everything.

Dan Costa: There were a lot of predictions made that in 2020 there would be fleets of autonomous cars out there on the road. A lot of those predictions have been walked back a little bit . How do you see the timeline developing? How far are we in this process?

Chris Urmson : I think none of us really understood just how hard this problem was. I famously said about my older son, “I’d like him to not have to get a driver’s license.” Turns out he’s going to be 15 and a half in I think two months, which means he can get a learner’s permit. So obviously we’re not quite there.

So the way we think about this at Aurora is our mission is to deliver the benefits of self-driving technology safely, quickly, broadly, and so we want to get to that point where we’re delivering that core benefit of it. But behind that, we feel this urgency to move, to get the technology into market, and start saving lives. And start making it easier to get around.

You’re right, people have walked back these timelines. I think there’s a lot of people who have limited experience in this space and they’re kind of guessing. And so now as we more deeply [to] understand it, I think within the next five years you’ll start to see kind of the early small scale deployments of this technology, [and] once we get to that it will start to scale relatively quickly. But this is a change that’s going to scale…over decades, not over…weeks.

Dan Costa: What are the obstacles you’re worried about? Are they technical obstacles? Are they legal obstacles? Are they moral obstacles, and it will take time to figure out how to program these algorithms to make the decisions we want them to make?

Chris Urmson : I think we’re going to face a sequence of challenges. I think the first one that unlocks it is actually getting the technology [to a point where] it’s good enough to be out there. And that’s really still hard. If you read some of the breathless headlines out there, you’d believe that the technology was done and you could buy it today. You can’t. So there’s a bunch of work there to both build the technology and convince ourselves it’s good enough.

As the technology reaches readiness, then we get into the mode of how do we most thoughtfully introduce this? As the technology moves from kind of the fantasy promise of what it can be, to the reality of what’s happening on the street, that’s where you see some of the bad events happen. And so there we need to kind of have done our job educating society, educating regulators, educating lawmakers around…why we’re building it.

These are some of the bumps we might see along the way, but if we get from here to the end state, we’re going to be much safer. We’re going to be much better. So kind of work with us through those and I think that will be the next phase of challenge for us.

Dan Costa: So when you talk about the moral complexity of building self-driving cars, it is more complicated than most engineering projects. People keep bringing up the trolley problem . What’s your take on solving that?

Chris Urmson : So the trolley problem is this philosophical question of imagine you have a trolley coming down a track. And it’s out of control. Let’s say there’s a nun on one branch and there’s a second branch where there’s a convict. You have the opportunity to throw the lever where you could divert it from hitting the nun to hitting the convict. What’s the right thing to do ? You can variate this. It’s three children versus an old man. It’s really a question that allows us to explore how do we value life and different aspects of life in our society.

Where it gets translated into the self-driving car space…you’re kind of in an inevitable collision. And the short answer is, there’s no correct answer, right? Philosophers have wrestled with this problem for centuries. It’s really, what do we as a society together believe is the right thing to do? The good thing is that self-driving cars should be much more alert. They’re going to be better defensive drivers, so it should rarely happen. I don’t know if in your lifetime you’ve ever had to pick between crashing into the wall or crashing into a person on the road.

Dan Costa: Most people don’t have to think those things out. And we fall back on human error. You can make the wrong decision. You can do the wrong thing and bad things will happen. And you’re only so responsible for making that mistake.

Chris Urmson : Agreed. But also, you live with the consequences, right? I think this is the part that people miss. One, people almost never have this happen. Self-driving cars will have it happen even less. The first premise in this is [that] people do the right thing. There have been studies that show in those kind of instantaneous events, it never gets to reasoning about which life is more valuable. It is an instantaneous reaction. Then the person that made that decision has to live with the consequences for the rest of their life.

I think that’s truly terrible. So the way I think about this is, let’s make it basically not happen. And then let’s describe what the outcome might be. We might say that the right thing is the car will work hardest to avoid vulnerable road users. Pedestrians and cyclists. And then after that it will work the next hardest to avoid other vehicles on the road. And then after that it will worry about not hitting walls and buildings.

Then people can say “Well, I don’t want to ride in that car.” Or they’re like, “Okay, I can live with that.” And particularly knowing that it’s basically not a risk and move on. We can propose that as the people delivering the technology. And then over time this will turn into a societal conversation around, what is the preferred outcome here? But I think the most important thing is to not let [being] perfect [get in the way of having] something…incredible…out on the road.

Dan Costa: That’s a great point. I want to be respectful of your time and ask you the questions I ask everybody who comes on the show. Is there a technology trend that concerns you and that keeps you up at night?

Chris Urmson : I think one of the things I think about a lot is, and it came up on the panel this morning…is the kind of asymmetry of some technologies. The connected world, the Internet of Things, if something goes bad, it can have a profoundly broad impact. There isn’t kind of diversity in the ecosystem, and that means one kind of point failure can bring down a lot of technology and so as companies get larger and larger technology…footprints gets larger and larger and more homogeneous. How do we protect against that? How do we provide diversity and immunity into the technology?

Dan Costa: Is there a technology or a service that you use every day that still inspires wonder?

Chris Urmson : I think there’s a lot of that, right? I see it all around. I’m an engineer, and the more time I spend on things, the more it’s clear how complicated pretty much everything is. Whether it is the fact that the cell phone in my pocket allows me to be simultaneously looking up whatever fact while talking to my parents up in Canada. That’s incredible. The fact that I have a car in my driveway that came off the line one minute before the car after it and has little explosions going off under the hood. And for the next 15 years it’s just going to work! That’s incredible.

The fact we flew here on a plane, and this giant thing with a couple hundred people in it. It stays up in there. That’s awesome. There’s a lot of anxiety in society right now, and when you take a step back and kind of look at the magic of everyday life, […]

What do you think?

486 points
Upvote Downvote


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.





Electric vehicle sales expected to reach 2m this year

Electric vehicle sales expected to reach 2m this year

At Ford, “creating the building blocks for autonomous fleets”

At Ford, “creating the building blocks for autonomous fleets”