Uber CEO Travis Kalanick unveiled the company’s first Uber Freight vehicle yesterday in Pittsburgh. In a tweet, he showed off the first step in his company’s plan to connect shippers with truckers, extending its ride-sharing model to cargo.
On the surface this news seems to boost Uber’s ambitions to take over, and eventually automate, not just human transport but also freight and logistics. After all, who could forget the tiny, less-than-a-year-old self-driving truck startup it acquired for an extraordinary $680 million last summer, which is now at the center of its legal imbroglio with Google?
But an investigation into the events that followed the acquisition of that startup, Otto, suggests that Uber’s work on self-driving trucks is in fact flagging. At a hearing in the US District Court in San Francisco last Wednesday, Waymo accused former employee Anthony Levandowski and Uber of using Otto as a shell to conceal his swift move to the ride-sharing company. “Uber and Levandowski created a cover-up scheme for what they were doing,” Waymo’s lawyer told the court. “They concocted a story for public consumption.” In other words, Uber wasn’t particularly interested in trucking — it was far more invested in the procurement of its personnel.
If Waymo is right, the level of theatrics is epic. Otto launched in May of last year with a slick film of an autonomous truck rolling down a Nevada highway with no one behind the wheel. “We’re going to continue our testing with urgency,” cofounder Lior Ron told Backchannel at the time.
Uber acquired Otto a few short months later, and through the fall of 2016 the ride-hailing company kept on developing its autonomous trucks. But a spate of recent evidence suggests that even if Waymo’s accusation is wrong, Uber’s commitment to automated trucking was shallow, and today its self-driving semis are nowhere to be seen. (The truck featured in the photo Kalanick tweeted last night shows no evidence of self-driving gear.)
Federal records show that Otto’s 18-wheelers are covering fewer miles than ever before. Key engineers have been reassigned to its self-driving car program. And ambitious plans for pilot projects across the nation have been delayed or cancelled. Perhaps most telling, emails obtained under public records legislation suggest that one month before Otto emerged from stealth, Levandowski was already envisaging its future as an autonomous urban taxi service.
So even as Kalanick boasts about his company’s latest venture into trucking, a parallel, conflicting story is now emerging around the company’s ambitions for autonomous freight. Otto, the gem in Uber’s portfolio of acquisitions, was drifting away from robo-trucking before the world even knew its name.
On April 11, 2016, Levandowski, Otto, and Uber signed a joint defense agreement, a crucial step towards Uber acquiring the young company. At the time, Otto was just a handful of engineers, operating quietly out of Levandowski’s home in Palo Alto. Two weeks later, the executive director of the Nevada Governor’s Office of Economic Development, Steve Hill, sent an email to the state DMV:
“Just wanted to let you know that a company is interested in operating completely autonomously, shuttling passengers on a fixed route in Vegas somewhere around the end of this year….
The company is headed by Anthony Levandowski. All this is confidential.”
Bruce Breslow, director of the state’s Department of Business and Industry, responded with: “Sounds like the Google project Anthony has wanted to set up for the last four years.” But when Otto launched publicly a few weeks later, with its video of a driverless truck and a spate of press interviews, its stated emphasis was entirely on autonomous trucking — with not a word about “shuttling passengers.”
In August, Uber acquired Otto and its 90-odd employees, and CEO Travis Kalanick put Levandowski in charge of Uber’s entire self-driving program, covering passenger cars as well as deliveries and trucking.
But Otto did not continue truck testing in Nevada, as Ron had suggested, or even apply for a permit to do so. Instead, the company started work in September on an autonomous vehicle certification center in a Las Vegas suburb — a facility that could license Uber’s self-driving taxis, as well as Otto’s trucks, to operate in the state.
The engineer who helped craft Nevada’s self-driving car regulations also ended up blowing past them.
Elsewhere, Otto’s trucks continued to make appearances throughout the fall, US Department of Transportation (DOT) records reveal. All commercial motor vehicles are subject to random roadside inspections, and in October Otto’s trucks had four, out of a total of eight for the year. That month Otto’s trucks completed their first paid delivery: a cargo of Budweiser beer in Colorado that was accompanied by a police escort. “We fully intend to dive deeper: more types of highways, more weather conditions, more traffic patterns, and obviously more partnerships in the future,” cofounder Ron told The Verge. In November, an Otto truck performed several test runs in Ohio.
But then Uber took its foot off the gas. Otto had approached Texas in August about deploying its trucks there, a public records request revealed, but the company never followed up on its plans, says Darran Anderson, Director of Strategy and Innovation at the Texas DOT. “The chain of email contact stopped after the Uber acquisition,” he says. “I just assumed the Uber buyout broke the contact or changed their focus.”
The company also stopped pursuing testing in Ohio, according to Randy Cole, executive director of the Ohio Turnpike. “Our facility passed their expectations with flying colors… but we just don’t have a firm schedule for when testing is going to start back up again,” Cole says. No further demonstrations have taken place in Colorado, either. And all of the engineers named in the application for its testing facility in Nevada are still based in San Francisco, according to LinkedIn. Emails addressed to the facility bounce.
One source close to the company, who asked to remain anonymous, says that Uber has largely stopped development work on self-driving trucks. Though public evidence is patchy, signs of that shift can also be found in Uber’s court filings regarding its lidar technology — a key point of contention in the Waymo lawsuit.
For instance, James Haslim started working at Otto in May 2016. “Shortly after I joined Otto, my team began working on a lidar project intended for autonomous trucks…known as Spider,” he wrote in documents filed as part of the lawsuit. “Uber acquired Otto in August 2016, and since that time, I have been responsible for the technical development of Uber’s lidar sensors.”
But Spider was ill-suited to self-driving cars. It was a bulky device that might have weighed six times as much as a commercial lidar from market leader Velodyne. “While the large size and heavy weight of the Spider was less of a consideration for a semi-truck, the weight of Spider design, along with the additional components …required to mount a completed Spider onto the roof of a passenger vehicle, would have likely exceeded the rated payload of the roof of a Volvo XC90,” wrote Haslim. Uber uses XC90s for its self-driving car tests in San Francisco, Pittsburgh, and Arizona.
Instead, Haslim’s team began work on a lightweight lidar called Fuji, designed specifically for cars. “Fuji is a lidar sensor having two optical cavities…oriented at different vertical angles to capture the field of view necessary for applications in self-driving cars,” he told the court. Fuji is intended to work at low speeds of up to 35 miles per hour, which is fine for urban taxis but virtually useless for long-distance trucking on highways. One of Uber’s recent filings says it has “no plans to revive the abandoned Spider project.”
In other court documents, Uber described Otto as “a company co-founded by Anthony Levandowski, which originally focused on self-driving trucks.” It did not say what the company is focusing on now. By the start of April, Uber had decided to erase all mentions of Otto, noting on its website: “We’ve retired the Otto name and integrated all of our self-driving efforts into Uber ATG.”
The most telling sign, however, is more straightforward. Otto’s trucks are simply not covering the miles they did at their peak in October — or even back in stealth last May. According to records, the company has not had a single DOT inspection nationwide so far this year, suggesting a very low level of on-road activity.
For Uber, a company beleaguered on many fronts, self-driving trucks may have quickly started to seem like an untenable distraction. In January, Uber lost 200,000 customers in a weekend due to Kalanick’s perceived ties to President Trump. February brought accusations of sexual harassment, and a video of Kalanick berating an Uber driver. March saw an exodus of executives, and revealed Uber’s history of evading and deceiving regulators using its Greyball software — now the subject of a Department of Justice inquiry.
Things on the self-driving front look nearly as dire. During a brief experiment in San Francisco in December, Uber’s autonomous Volvos were spotted running red lights. In court documents, Uber admits that it has never successfully deployed any of its own lidar sensors, and that it fears being sidelined in “what potentially may be the most lucrative business in history.” Rivals GM and Waymo are both rolling out fleets of self-driving taxis hundreds strong. For Uber to expend time and effort on self-driving trucks today would seem foolish, to say the least.
“If Otto is still working on self-driving trucks, Travis is way dumber than I think he is,” says Stefan Seltz-Axmacher, CEO and founder of Starsky Robotics, a self-driving trucking company which is already hauling freight for money in Florida. “If Uber loses the self-driving taxi race, there isn’t going to be an Uber.”
Uber has every reason to focus its technological might on its core business of personal transportation. As if developing self-driving cars was not difficult enough, the company recently held a three-day conference, Uber Elevate, that explored what it calls “on-demand urban aviation” and what everyone else calls flying taxis. It announced partnerships with cities, property owners, and companies working to build and support electric aircraft capable of vertical take-off and landing.
Sebastian Thrun Defends Flying Cars to Me
The CEO of Kitty Hawk, the Larry Page-funded personal airborne vehicle company, explains why this isn’t the stupidest…
“On-demand aviation has the potential to radically improve urban mobility, giving people back time lost in their daily commutes,” said Jeff Holden, Uber’s Chief Product Officer. “We think we can start this for roughly the cost of UberX.”
The company insists that its new passion for aviation does not mean Otto’s dream of self-driving trucks is over. An Uber spokesperson told Backchannel: “We have two products and two product-dedicated teams that collaborate on cars and trucks under the same leadership. Our truck team is hard at work improving the technology and we’re looking forward to demonstrating our progress soon.”
Uber says it will publicly demonstrate its first flying taxi service in 2020. No date has been given for its self-driving trucks to hit the roads commercially.