Is Uber's Self-Driving Program Veering Off Track?

Image
Andrew Harrer/Bloomberg News

A self-driving car flipped on its side seems guaranteed to make humans think twice about riding with a robot.

Never mind that Uber's accident last weekend in Tempe, Arizona, was the result of a human failing to yield and smacking into the autonomous Volvo. Despite the lack of serious injuries, that image alone raises questions about both our acceptance of self-driving tech as well as Uber's rabid rush to bring this new age of mobility to life.

"We may lose 35,000 people a year to traffic deaths, but if 10 people are killed by autonomous cars, we'll be freaking out," said Karl Brauer, director of content at Autotrader and Kelley Blue Book.

That it happened to Uber seems to be a bit of bad luck blended with hubris, added Brauer, noting that in some 2.5 million miles and years of testing the most that has happened to Google's autonomous cars are low-speed bumper dings.



RELATED: Rep. Lipinski — time to think about how to regulate self-driving vehicles

"Google has taken the slow and steady approach, and they're the leader," he said. "Uber is a company that leverages technology [to connect drivers with riders], but they're not a tech company. They seem to be frantically transforming themselves into something they're not by nature."

Uber responded to a request for comment with confirmation that its cars will be back on the road March 28 in Arizona and Pittsburgh.

John Krafcik, CEO of Alphabet's veteran self-driving project, Waymo, meanwhile tweeted out a photo Monday of a self-driving car testing in a winter wonderland. "Snow practice!" he wrote, with the location identified as South Lake Tahoe.

RELATED: Daimler, Uber to partner on self-driving cars

Testing in inclement weather is considered the Achilles heel of autonomous vehicles because their sensors can't see the road and must rely on highly detailed 3D maps. Waymo, started as Google's self-driving program, seems to be upping the ante just as a rival is struggling.

Uber's battle against its rival's seven-year-old program is "almost existential," says Gartner auto analyst Mike Ramsey. "They're trying so hard, but this can't be done overnight even if you hire the smartest people in the world."

Ramsey said accidents are part of any experimental programs, and there's no reason to think Uber would be exempt. But the Arizona incident puts a spotlight not only on the technological hurdles that remain but also the psychological chasm that still exists between today's deadly tolls and a potentially safer but autonomous future.

"There's already a huge issue when you hand over the driving to someone else," he said.

RELATED: Self-driving truck hauls freight in Ohio as state unveils investment to aid research

All told, more than 30 companies large and small — from Audi to Waymo — are working on the hardware and software components of self-driving cars. Many companies point to 2020 or 2021 as an internal goal for creating a commercially viable autonomous car, one that likely would first appear as part of a ride-hailing service.

But that goal is contingent on other factors, including cohesive federal rules on testing and deployment, agreement on insurance issues and infrastructure needs such as clear lane markings. In that sense, Uber may have egg on its face today, but there still is time to determine the big winners in this lucrative mobility revolution.

"This is transformational stuff, but I'm skeptical it'll evolve fast," Ramsey said.

In some ways, Uber's headlong rush into the self-driving car race echoes Icarus, whose waxen wings melted as he soared toward the sun.

Despite having no expertise in the space, two years ago the ride-hailing giant decided it would take on tech companies and automakers alike in the complex quest for autonomous vehicles, a goal that would eliminate a big cost to Uber rides — the driver.

Uber raided Carnegie Mellon University for its best roboticists, spent $670 million to buy autonomous truck company Otto and, last August, proudly started picking up Pittsburgh passengers in robot-driven cars.

But fast forward, and the wax on Uber's wings appears to be melting.

Key members of its self-driving car division have left, and a recent report on tech website Recode described the company's autonomous car team as being in a "mini-Civil War." Waymo is suing Uber, claiming that Otto's key sensor technology was stolen by Otto founder and ex-Google employee Anthony Levandowski. And then comes the embarrassing accident in Arizona.

RELATED: Uber says trade theft case is between Otto chief and Google

Even the path to that crash was bumpy: Hours after Uber started testing its self-driving cars in December, the California Department of Motor Vehicles told it to halt because it hadn't received the proper permit. Instead of complying, Uber sent its fleet to Arizona.

California's DMV, which granted Uber approval for testing earlier this month, is "contacting Uber and working with our partners in law enforcement to find out more about the crash involving a self-driving vehicle in Arizona," said Jessica Gonzalez, a spokeswoman for the agency.

Uber, which was founded in 2009 and is valued at close to $70 billion, became the biggest startup in history thanks in part to playing hardball with drivers, municipalities and state officials.

Its contentious relationship with its drivers is well-documented, with many filing lawsuits to be considered employees in order to receive better pay and some benefits. But those have paled next to damaging allegations about a male-chauvinist company culture that resulted in CEO Travis Kalanick going on the hunt for a chief operating officer while former U.S. attorney general Eric Holder conducts an internal investigation.

Last month, Uber admitted that it would stop using a technology called Greyball, first reported by The New York Times, to deceive city officials who wanted to take rides in an attempt to regulate the service.

Uber ultimately has little choice but to continue its feverish quest for proprietary self-driving tech. Although the private company's finances are a mystery, industry experts question whether the company is subsidizing rides to grow market share, much as it did recently in a failed attempt to break into the Didi Chuxing-controlled Chinese market.

Eliminating the driver would improve Uber's business model, but significantly less so if that self-driving technology had to be purchased from a competitor.

"Uber remains a huge player [in autonomous car research] just given that huge valuation alone," Brauer said. "But they're also not the only ones working on this stuff."

There is also, for Brauer, a strong sense of comeuppance in the Arizona accident.

"They've had this irreverent, shoot from the hip attitude about so many things, saying they'll just hire their way into this or buy their way into it or they'll ignore the [state] rules," he said. "It just makes me think that if you're Travis, and you're 40 and are worth tens of billions, maybe you have an overblown sense of what you can do. We will see."