PITTSBURGH — On our last trip to the Steel City, we took a ride in one of Uber’s self-driving vehicles. It didn’t go so well.
More than four years later, we returned to Pittsburgh, curious to see how things have changed in the world of autonomous vehicles (AVs).
We spotted a number of self-driving cars roaming around city streets on test drives. And much technological progress has been made by a bevy of companies worldwide that last year raised record funding from investors.
But the promise of self-driving technology still appears years, perhaps decades away.
Industry insiders remain bullish.
“We have seen a lot of players come and go that have all kinds of shiny tech and demos — yet self-driving remains this elusive problem,” said Dima Kislovskiy, vice president of integration at Aurora. “But we continue to believe in the tremendous benefit it can deliver, now more than ever.”
This week we visited Aurora’s growing footprint in Pittsburgh, a robotics hub that is home to other self-driving companies such as Argo AI and has a long history of AV research at local universities including Carnegie Mellon, where Aurora CEO Chris Urmson completed a PhD in robotics.
Aurora always had a presence in Pittsburgh since launching in 2017 but recently established its headquarters here following its acquisition of Uber’s self-driving unit, which had operations in the city.
Uber invested $400 million in Aurora as part of that deal, which valued the combined company at $10 billion. Aurora then went public in November via a SPAC merger.
Both transactions equipped the 1,600-person company with serious cash and talent to fuel its commercialization plans — helpful for a pre-revenue company that posted a net loss of $76.8 million in the first quarter.
Aurora, which is also backed by Amazon, is building a “driver-as-a-service” model that provides self-driving technology to an external fleet owner or operator.
It is starting with an autonomous trucking product, Aurora Horizon, that is being tested in Texas. Those trucks are still under the supervision of vehicle operators. Horizon is planned to launch in late 2023 with around 20 vehicles. Aurora’s partners include Seattle-area trucking giant Paccar, Volvo, FedEx, and Werner Enterprises.
Trucking seems like a logical application for self-driving innovation. Human drivers are limited in the number of consecutive hours they can drive; a robot doesn’t need sleep. It’s also a job of “stress, physical deprivation, and loneliness,” The New York Times reported in a story about the driver shortage.
The recent supply chain chaos and rising inflation also highlight the efficiency benefits of self-driving trucks. And Aurora believes its tech can help “make the road safer,” said Kislovskiy.
“We’re seeing all of these things that really make clear the need for this technology,” he added.
Aurora’s secret sauce may lie within its “FirstLight Lidar” tech which it says is better than traditional lidar due to long-range sensing capabilities. Here’s a recent video showing how Aurora Driver can navigate around construction zones. The company said it has reduced required human intervention by 18-fold during tests in Texas.
The idea is to use learnings from self-driving trucks and apply them to “robotaxis,” using partnerships with companies such as Uber and Toyota. Aurora in March unveiled its first fleet of custom-built Toyota Sienna vehicles, which can operate at highway speeds of up to 70 MPH and are being tested in the Dallas-Fort Worth area. The company plans to launch the ride-hailing product, Aurora Connect, in 2024.
“You don’t have to take on the whole problem space at once,” said Kislovskiy, who spent two years at Uber before joining Aurora in 2018. “We’re building our core technology to serve both of these use cases.”
There are many hurdles to clear before Aurora’s business can take off. It faces not only technological, regulatory, and safety challenges, but also competition from well-funded rivals including Alphabet’s Waymo and General Motors’ Cruise arm. Tesla and Mobileye are also making progress on their own “robo-taxi” rollouts.
Urmson told analysts this week that there is room for more than one winner in a U.S. market he estimates at more than $1 trillion.
“It’s extremely early innings,” said Urmson. “We are somewhere in the pre-Model T phase of self-driving.”
Aurora sees advantages with betting on trucking first — “very clear demand” and stronger economics and operational scale opportunity compared to ride-hailing, Urmson said.
Amazon, which disclosed a 5% stake in Aurora earlier this year, might be a potential Aurora customer if self-driving trucks ever take off, given the Seattle tech giant’s logistics needs.
“Our first application is one that we believe is highly relevant to what Amazon is doing,” Kislovskiy said.
The plan is to generate revenue from trucking and use it to enter the ride-hailing market, where it could differentiate with its connection to Uber.
“We really want to make sure that we’re focusing on delivering a real product that has real commercial appeal, and has a real societal benefit,” said Kislovskiy. “Because if not, then we’re just an R&D lab. What is this all for if we don’t actually do that?”
Urmson previously led Google’s self-driving arm before it turned into Waymo. He started Aurora with Sterling Anderson, former head of Tesla Autopilot, and Drew Bagnell, a former leader of Uber’s self-driving team. Some of the company’s early investors include Greylock Partners, Sequoia Capital, and T. Rowe Price.
“Progress in self-driving technology development is a continuum,” Urmson wrote in the shareholders letter. “Each incremental advancement represents a meaningful step towards our objective of commercial deployment at scale.”