Fully autonomous vehicles are more commonplace than you might expect as cars and big trucks gain the ability to drive safely on our streets and highways.
As one of the pioneers in the field, Chris Urmson was there almost from the start, competing in the landmark DARPA Grand Challenge before leading Google’s self-driving car initiative. Now he’s realizing that vision as CEO of Aurora, which develops self-driving technology for auto and truck manufacturers.
In this episode of the GeekWire podcast, we play highlights from our conversation with Urmson at the last GeekWire Summit.
Listen below or subscribe to any podcast app and read on for excerpts.
Aurora’s Approach: We founded Aurora about six years ago with a mission to deliver the benefits of self-driving technology safely, quickly, and widely. We build the driver technology. We don’t want to build trucks or cars. There are people who are really good at it. We don’t want to build Uber, FedEx, or any other carrier. We want to develop the powertrain technology to power these companies.
Underlying technology: The driver we built uses a combination of sensors: LIDAR, radar, cameras, our special high-resolution maps. We have our proprietary FirstLight LIDAR, which allows us to see further than others. And we have a lot of computers on board. It is this common architecture, hardware and software that runs everything from the Toyota Sienna to PACCAR and Volvo vehicles.
Current status of operation: Today in Texas we have trucks on the streets pulling loads for customers every day, with people on board, with operators on board. … The vast majority of the time it drives itself. And that’s on the freeway. And it’s a very smooth, capable driver at this point. It does the things you would expect other drivers to do. So if it sees vehicles trying to merge, it makes room and moves over. If it sees a vehicle stuck at the side of the road, it will slow down, back away, and do all the things a good, conscientious driver would do.
Why start with big rigs? For a number of reasons:
- There is just an incredible need for this technology in the US. Today we are short of 80,000 drivers; We anticipate a 160,000 driver shortfall by the end of the decade. This is one of the factors contributing to the supply chain challenge.
- Security is very important. There are around half a million serious truck accidents every year. We can do something about that.
- As a company, the economic opportunities are even better. Trucking is a $700 billion industry in the US today; Ride-hailing is a $35 billion industry. And so, from an addressable market, it’s profound.
- If we consider the possibility of building and scaling the business, the unit economy is stronger. Put simply, we pay truck drivers three times what we pay ride-hailing drivers. So if you’re thinking about adopting a technology, it makes it easier to start scaling the business and moving into other areas.
- Finally, from a technological point of view, we expect to be able to scale faster. If you think about how a mile of freeway in Texas compares to a mile of freeway in Minnesota or a mile of freeway in California, they’re all basically the same. Whereas at an intersection in San Francisco you walk five blocks and it’s different people, different behavior, different geometry.
What do you say to people concerned about security? First, this is a new technology. It’s very rational and reasonable to have concerns and questions and really want to understand them better. This is perfectly normal and healthy.
Security is at the core of the company’s DNA. For this reason, we are so transparent about how we ensure security in the company. We have shared our framework for how we will convince ourselves and others that the vehicle is safe.
The technology is magical in a way. It can look in all directions at once. And it doesn’t have the human reaction of the foveation that happens. … There is an incredible opportunity for safety here.
If you could drive I-5 from the Bay Area to Seattle yourself or have the Aurora Driver drive it, what would you choose for safety?
We’re not quite there yet with the Aurora Driver. But we’re making really good progress. There’s a lot of times I would certainly trust him. Where we’ve come with the Aurora driver is increasing reliability… but we’re not quite there yet. Otherwise we would get by today without a driver.
What’s next: We are working to have the feature ready by the end of Q1. At this point, the Aurora driver fulfills everything it needs to be part of any product anywhere in the world. But it still doesn’t quite do it well enough. We are working toward the end of next year to be ready.
Watch the full episode to learn more, including the parallels between Aurora and Microsoft, plus Urmson’s thoughts on Amazon’s acquisition of Zoox and Amazon’s investment in Aurora.
Listen above or subscribe to GeekWire on Apple Podcasts, Google Podcasts, Spotify, or wherever you hear it.