×

Inside Uber’s self-driving car mess

The company’s self-driving division is in a “mini civil war,” according to insiders.

For Uber, the embattled ride-hail juggernaut now valued at $70 billion, much of the company's future is resting on what happens this week.

But it's not just the meetings that CEO Travis Kalanick is having with potential candidates to become its first chief operating officer — an appointment focused on fixing the massive controversies that have washed over the ride-hailing company around its dysfunctional management.

He's also gathered leaders from Uber's self-driving division in Pittsburgh and its counterpart in San Francisco for a critical summit that started on Monday. While this isn't the first meeting between the leaders of its kind, sources say this one has been aimed at hashing out issues of leadership, the entity's technological progress and questions about what the priority should be for the Advanced Technology Group (ATG), the name of Uber's self-driving unit.

Travis Kalanick, co-founder and chief executive officer of Uber Technologies Inc.
Udit Kulshrestha | Bloomberg | Getty Images
Travis Kalanick, co-founder and chief executive officer of Uber Technologies Inc.

The list of attendees at this self-driving summit has included engineers both old and new — many of whom were either poached from Carnegie Mellon or joined in August of last year when Uber acquired self-driving trucking startup Otto.

David Stager, a Carnegie Mellon alum and top dog at the Pittsburgh facilities, has been joined by Otto co-founders Don Burnette and Lior Ron and at least 50 others.

Uber's future depends greatly on solving self-driving. It's what will keep the ride-hail company relevant as more automakers produce their own autonomous vehicles. But taking drivers out of the equation would also increase the company's profits: Self-driving cars give Uber 100 percent of the fare, the company would no longer have to subsidize driver pay and the cars can run nearly 24 hours a day.

But the company's autonomous efforts are in turmoil. According to extensive interviews Recode conducted with former and current employees at the self-driving effort, many think it is at a technological standstill and plagued by significant internal tension, especially among its executive leadership.

More from Recode:
WTF is happening at Uber?
Uber says it just had its best U.S. week ever — despite its executive turmoil and image problems
Uber's vice president of mapping, Brian McClendon, has resigned

The issues have included a wave of key talent departures and problematic demos. At least 20 of the company's engineers have quit since November. One source says a "mini civil war" has broken out between those who joined Otto in search of the independence of a startup, and those who joined Uber's ATG with ambitions to solidify the company's place in the future of transportation.

Many of those issues and the resulting questions can be traced back to when Uber acquired Otto, several sources said. As part of the acquisition, Kalanick put its founder and CEO Anthony Levandowski in charge of all of its autonomous efforts.

Uber says that it's normal for an entity that was founded two years ago as of January 2017 to see this level of attrition, particularly as the company recently paid out its employee bonuses. However, the departures began as early as November 2016. Additionally, a company spokesperson said Uber's ATG has seen fewer departures than the overall company and has hired more people than have left since the start of the year.

Now, Levandowski is at the center of a major lawsuit his former employer Alphabet has filed against Uber alleging he stole key intellectual property when he left to create Otto. The lawsuit, which claims Levandowski had a long relationship with Uber before leaving Alphabet, could severely handicap the company's autonomous efforts if a judge rules against Uber.

It's unlikely all these issues will be remedied within a week, particularly since Kalanick's focus is divided between the company's autonomous efforts and its recent public scandals.

But sources said the controversial CEO has repeatedly held meetings with the company's human resources representatives, Levandowski and Eric Meyhofer, a top engineer who joined the Pittsburgh team more than two years ago, just in the first two days of the week.

A 'brother from another mother' arrives

Still, coming up with a solution that would make all stakeholders satisfied won't be easy, particularly since the circumstances around the Otto acquisition itself are in question.

Reconstructed timeline based on our reporting, previous reports and Waymo's lawsuit:

2012: Anthony Levandowski and Travis Kalanick first meet at a TED Talk.

2015: Alphabet alleges that Levandowski met with Uber's vice presidents of maps, Brian McClendon.

December 2015: Levandowski allegedly downloads proprietary data from Alphabet's self-driving files, according to the suit.

January 2016: Levandowski leaves Alphabet's car group, now called Waymo, and sets up a new company called 280 Systems, which eventually becomes Otto. Alphabet alleges that Levandowski was later seen at Uber headquarters.

February 1, 2016: Otto is officially formed.

Between February and March: Levandowski begins appearing at the company's Pittsburgh headquarters. Kalanick hires him as a consultant, according to sources.

May 16, 2016: Otto launches out of stealth.

Spring of 2016: Kalanick brings up the possibility of acquiring Otto to Levandowski.

August 2016: The original heads of Uber's self-driving division, including former director John Bares, planned to have cars on the road by this time.

August 18, 2016: Uber acquires Otto.

Around the same time Otto was formed in February 2016, Uber hired Levandowski as a consultant for its self-driving efforts, Recode first reported. The timing is unusual, given that Levandowski was starting his own self-driving company.

Originally, the ATG was working toward getting its cars on public roads by August 2016. In the months leading up to that deadline, Kalanick determined that the team, mostly comprised of Carnegie Mellon roboticists and engineers, wouldn't be able to come through on their promise, because the technology was just not ready. So, Kalanick turned to Levandowski — who, according to reports, he met in 2012.

Per Levandowski's advice, the deadline was shifted to September 2016. But while the ATG ramped up its self-driving efforts, so too did Levandowski and his co-founders Ron, Burnette and Claire Delaunay — but they were working on trucks. Otto launched out of stealth in May, about four months after Levandowski formed the company and began consulting for Uber, at which point the company's autonomous trucks were already being tested on public highways.

It's not entirely clear why someone starting his own self-driving company would help advise another company's self-driving efforts. But Levandowski's long relationship with Kalanick might have something to do with it.

Anthony Levandowski, Otto Co-founder and VP of Engineering at Uber.
Angela Merendino | AFP | Getty Images
Anthony Levandowski, Otto Co-founder and VP of Engineering at Uber.

According to one report, Kalanick and Levandowski first met at a Ted Talk in 2012, where Kalanick apparently first broached the idea of using some of Google's technology. It wasn't until the spring of 2016, however, that the duo began discussing merging Otto into Uber.

By the time Uber had decided to publicly announce its decision to purchase Otto in August, Kalanick said he felt Levandowski was "like a brother from another mother."

However, if allegations in Alphabet's lawsuit against Uber are true, those acquisition conversations may have actually begun as early as the summer of 2015 — approximately six months before Levandowski even left Alphabet.

"Mr. Levandowski had previously told me, in or around the summer of 2015, that he had talked with Brian McClendon, an Uber executive involved with their self-driving car project," a sworn testimony from Waymo engineer Pier-Yvez Droz reads. "He told me that it would be nice to create a new self-driving car startup and that Uber would be interested in buying the team responsible for the LiDAR we were developing at Google."

Those conversations continued into early 2016.

"Waymo is informed and believes that Mr. Levandowski attended meetings with high-level executives at Uber's headquarters in San Francisco on January 14, 2016," the complaint reads.

That lawsuit came just eight months after the acquisition of Otto, adding to the growing malaise among rank and file engineers at the ATG.

In its complaint, Waymo — now the name for Alphabet's self-driving efforts — alleges Levandowski stole the design of its laser radars, known as lidars. They're used to spot and measure distances to other objects so cars can avoid collisions. To prove to employees that the claims are, as an Uber spokesperson said, baseless, Kalanick and Levandowski addressed the ATG at an all-hands meeting the day after the complaint was filed.

The laser team took apart the system to show how it was built in an effort to stave off concern. But it was unclear how employees who didn't work at Google would know whether it looked like Google's design.

No one knows how the company will move forward in the face of separate allegations of sexual harassment and stealing trade secrets. Though the lawsuit is only one in a series of issues at Uber's self-driving unit, it's the one that could bring the entity to its knees.

A talent exodus begins

It's not clear precisely when Uber executives began discussing buying Otto, but what happened shortly after after was worrying. Within months of the company acquiring Levandowki's startup, a wave of the group's top engineers — many of whom had been poached from Carnegie Mellon University's famed robotics institute — left.

The exodus included Peter Rander, who has since started his own autonomous startup called Argo.ai; Brett Browning, who has joined Rander's startup as the vice president of robotics; and Drew Bagnell, the head of computer vision at Uber who has now joined former Google self-driving CTO Chris Urmson's new autonomous startup, Aurora. A few months later, Raffi Krikorian — the former vice president of engineering at Twitter who was hired by Uber in 2015 to lead its self-driving division— also left the company.

In addition to four of the top engineers departing, there are at least 15 others who have left, Recode has confirmed. Many of those who've stepped down from their roles at the company have joined Rander and Browning at Argo. That includes the ATG's former director of finance Daniel Beaven and a slew of engineers that worked on Uber's mapping team: Randy Warner, Dale Lord, Tyler Krampe, Al Costa, Robert Keelan and X Xinjilefu. At least six others have joined Urmson and Bagnell at Aurora, according to sources.

The departures have continued this month. On March 3, the security expert Charlie Miller — famed for hacking into a Jeep remotely — announced he was leaving the company to join Didi's self-driving effort. And, only a day into the self-driving summit at Uber's headquarters this week, the company's longtime technical program manager Pam Cardona also quit.

High turnover is not in and of itself uncommon at startups, but it is obviously a major warning sign here. More importantly, retaining talent is crucial for Uber as it attempts the difficult task of being a leader in the ever-competitive market for on-demand, self-driving cars.

There are only so many engineers who have any expertise in this area and now, more than ever, demand for these highly skilled engineers has sky-rocketed. Traditional automakers, big tech giants and small, well-funded startups are scooping up all the talent they can through acquisitions and mergers.

Pilot models of the Uber self-driving car is displayed at the Uber Advanced Technologies Center on September 13, 2016 in Pittsburgh, Pennsylvania.
Angelo Merendino | AFP | Getty Images
Pilot models of the Uber self-driving car is displayed at the Uber Advanced Technologies Center on September 13, 2016 in Pittsburgh, Pennsylvania.

So what accounts for the troublesome exodus of top talent at Uber's ATG?

For one, adding to confusion about the chronology of events around the merger, higher-ups who were keyed in and informed about the impending acquisition of Otto were not told that Levandowski would effectively become their boss until shortly before it went through.

Since then, there has been a substantial shift in ATG's management structure — much of which was done without any consultation from the department's original leadership — that put a number of Otto's team at the top of the hierarchy, with demotions for some of those who started with ATG.

For instance, Stager, who has been with ATG since its inception, works alongside Otto co-founder Don Burnette as the autonomy tech lead, according to an email Recodeobtained. And John Bares, a CMU roboticist who was once the director of and an instrumental part of the ATG, has been demoted to the head of test operations.

Without warning, the company also sent Paw Anderson, formerly the head of business infrastructure services at Uber's headquarters in San Francisco, to be the ATG's senior director of engineering. ATG staffers weren't told by Levandowski or Kalanick that Anderson would be joining the self-driving arm and did not become aware of it until he showed up at the company's Pittsburgh facilities one day, according to sources.

The Pittsburgh team is expected to assert their leadership at this week's summit in San Francisco. Whatever the outcome of the series of meetings the company's self-driving arm will be having this week, it's clear something needs to be done and fast. With mounting competition from established and new players — like Rander's Argo, which Ford has now acquired a majority stake in — Uber can't afford this level of attrition.

That's to say, with every engineer that defects, Uber is feeding the fire of its competitors, which are growing by the day, both big and small.

Running red lights

For many of those engineers that left, there was growing concern around Uber prioritizing demonstrating its self-driving car progress in public, rather than more quietly developing and perfecting that technology until it is completely ready for debut.

The shift to show off Uber's progress to the world has come from Levandowski, insiders say. Rather than dedicating the unit's resources to meeting long-term goals, many said he prefers putting on public demonstrations to convince people that Uber is ahead of its competitors.

An Uber spokesperson said that real world tests are critical to the success of the technology and the company has always had plans to launch pilots.

But sources say that laser-focus on public demonstrations led to the rushed launch of Uber's self-driving pilot in San Francisco, as well as a haphazard demo of Otto's autonomous trucks performing a beer delivery in Colorado. Both are examples of the problems plaguing the unit.

After the September launch of semi-autonomous vehicles in Pittsburgh — which was already part of the original plan — Levandowski rushed another demonstration a month later.

One of Otto's trucks drove itself 120 miles down a public highway in Colorado to deliver 45,000 beers as part of a deal with Anheuser-Busch. This took place only eight months after Otto was founded, so it came as a surprise to many that the company was able to show off its technology so early in its existence.

According to an Uber spokesperson the Otto team worked closely with Colorado officials and ran extensive tests with the trucks before meeting an internally set safety criteria and launching the demo. Indeed, as BuzzFeed News reported, the company completed five tests without human intervention.

However, the team was able to put on a show because they cobbled together unfinished systems that were "thrown together," according to a source.

To be fair, it's not exactly uncommon for an early-stage startup to demo technology when it's not yet fully formed. When investors looked at virtual reality headset Oculus, for instance, it was still duct-taped together. Even Cruise — a self-driving startup acquired by General Motors — demoed unfinished technology to potential investors.

But the stakes are higher when it comes to self-driving vehicles, which have the potential to cause fatal accidents. Even though Otto's truck was driving itself on a public highway alongside other human drivers, the computers were crashing every few hours, these people said.

"A huge amount of luck went into that demo," one said.

Fortunately, the truck was accompanied by a motorcade of state troopers during its demo. But for some of Uber's top engineers, luck was simply not good enough.

Sources say that after the Otto demo, some ATG staffers became concerned about the ethics of fielding unsafe systems and rolling them out onto public roads. Shortly after the October demo, Rander, Browning and Bagnell left the company.

Those departures came at a critical time for the company.

Just two weeks after the engineers left, Uber began its second commercial test of its autonomous cars — this time in San Francisco.

Even to the public, it was clear the rollout of these Volvos was a logistical disaster. On December 14, or the first day of Uber's commercial test, one of the cars ran a red light. While Uber maintains it was due to human error, a damning New York Times reportrevealed the internal logs showed the system failed to recognize the traffic light, a significant technological issue to say the least.

But with nascent semi-autonomous technology — which means it's not supposed to completely drive itself — safety drivers have to be at the ready to take over at a moment's notice. A skilled and trained safety driver would've prevented this, several people contend.

But, "we don't have those [types of] drivers in San Francisco," one source familiar with the matter said.

That wasn't the only hole in the company's plans to roll out its cars in its hometown. The scene ahead of the launch was just as messy, sources say.

As early as September, Levandowski began emailing with the California Department of Motor Vehicles, arguing that the company didn't need to obtain the $150 permit required of all autonomous vehicle companies to operate semi-autonomous cars on public roads. The DMV argued that it did, but Levandowski informed the department Uber would continue with its plans anyway.

By late October, it became clear the operations team in San Francisco was struggling to prepare for the public launch of the company's on-demand autonomous cars, according to insiders.

It was then that a number of software engineers and safety drivers from the Pittsburgh arm of the ATG were sent to California to help the San Francisco team, which had problems performing certain critical tasks, according to sources.

It's not unusual for cross-functional teams to help out on getting pilots up and running, an Uber spokesperson contended. This was also the case when the company demonstrated its technology in Arizona recently.

For the San Francisco pilot, the engineers were sent to help with tasks such as determining what data was missing — like where traffic lights are located — and what the vehicle was not seeing, as well as things like how to onboard new code.

Part of the reason the California launch wasn't as seamless as the Pittsburgh launch was because Uber had originally charted out parts of Arizona with plans to begin its second self-driving test in the state by the winter of 2016, according to several sources.

After the red-light disaster, the California DMV put its foot down and revoked the company's licenses, forcing Uber to ship its vehicles to Arizona, the site of Uber's original testing plan. Levandowski had insisted on moving it to San Francisco since Arizona wasn't flashy enough, sources said.

Uber, no stranger to local regulatory battles, has since applied for and received the proper permits to operate in California. In the coming weeks, the company expects to roll out a number of its self-driving Volvos in San Francisco, according to internal documents Recode obtained.

The progress of the technology is "not great"

The concerns of these engineers, many of whom have departed, have some merit, according to other internal documents Recode obtained. While Uber's self-driving cars are putting on more autonomous miles every week, the technology has not improved as steadily.

In the months since these top engineers have left, the 43 self-driving cars Uber had on the roads as of March 8 saw little progress when it came to decreasing the frequency of disengagements — or how many times a safety driver would have to take back control of the system.

Uber's cars had driven more than 20,300 miles in autonomous mode in just that seven-day period ending on March 8, according to the documents. That's a big increase from earlier in the year, which reflected the increase in the number of autonomous cars on the road.

An Autonomous trucking start-up Otto vehicle is shown during an announcing event in Concord, California, U.S. August 4, 2016.
Alexandria Sage | Reuters
An Autonomous trucking start-up Otto vehicle is shown during an announcing event in Concord, California, U.S. August 4, 2016.

But, while the cars saw some minimal improvement week over week, they weren't getting increasingly safer at a steady pace and drivers were still having to take back control an average of 10 times for every eight miles driven.

That's also true of what the company calls "bad experiences," or how many times the robot car abruptly decelerated, braked or jerked around.

For example, the cars had more "bad experiences" during the week ending on March 8 than in January. The miles driven between instances the car did something it wasn't supposed to do has dropped from more than four miles in January to less than two miles as of last week.

In other words, cars only went an average of two miles without some sort of sudden movement or otherwise bad experience.

These incidents have less to do with safety and more to do with the total rider experience and how smoothly the car is able to drive itself.

In an email to staffers at the ATG, the company describes the total rider experience — which is measured by a combination of how many times the safety driver had to take over and how many bad experiences the car encountered — on one route in Arizona as "not great."

The big caveat here is that Uber just started testing its autonomous cars in Arizona and had a better track record around Arizona State University, which the company mapped out as early as 2015.

Even in Pittsburgh, where Uber's autonomous cars have a few months on those in Arizona, the company's progress is unsteady.

"The slightly sore spot is Pittsburgh's routes," a weekly email on internal metrics ATG staffers received read. "They showed lower than average [miles per harmful event.]"

An Uber spokesperson says the ATG team takes great care to ensure the software the cars are running is safe and in fact has set up a protocol to ground all cars if questions about the stability of the system arise — which Uber claims is uncommon in the industry.

Truck versus car

Along with these test mishaps and the slow progress of the technology, there has also been a clear tension between the ATG team, which is focused on cars, and the Otto team, which is focused on trucks. The two sides have argued over which group should be the business priority. That's partly because Kalanick hasn't given clear direction. Sources say they hope that will become clear after this week's summit.

According to a spokesperson, however, Uber sees both cars and trucks as a critical part of its mission to provide reliable transportation.

Engineers on both the Otto and the Pittsburgh teams feel the acquisition has stunted their development, according to sources. Otto employees, many of whom joined after leaving larger companies like Google and Apple in search of the freedom of a startup, were promised they would remain independent and would continue working on the trucks. But that hasn't been the case.

With much of Otto's leadership focusing on the development of the autonomous car technology and heading up all of the company's autonomous efforts, the trucks have become a secondary focus.

"The Otto team really doesn't like the setup and feels like [Levandowski] pulled one on them," one source said.

For the Pittsburgh team, Levandowski's focus on public demos — which require both time and resources — has set the team back in its pursuit of launching a network of driverless cars, insiders say.

There has, however, been commercial progress on both fronts. Already, the team focused on Uber's marketplace for cargo shipments — called UberFreight — has approached potential customers. On the car front, the company has struck a second self-driving car partnership with Mercedes Benz after buying 100 Volvos to outfit with its own tech.

Lack of Kalanick

Notably, the all hands meeting discussing the Waymo lawsuit was on of the rare times Kalanick has addressed the entire ATG team, much to the dismay of its top engineers. It wasn't until December, when top staffers began to leave, that Kalanick spent significant time with the whole ATG team in Pittsburgh.

That may change, given Kalanick's hands-on approach to this week's meetings. While self-driving technology won't solve Uber's financial woes in the immediate term, it will be key to Uber's relevance in the next decade or so.

If so, Kalanick can finally deliver on the points he made in August.

"The minute it was clear to us that our friends in Mountain View were going to be getting in the ride-sharing space, we needed to make sure there is an alternative [self-driving car]," he said, referring to Google, in an interview. "Because if there is not, we're not going to have any business."

By Johana Bhuiyan, Recode.net.

CNBC's parent NBCUniversal is an investor in Recode's parent Vox, and the companies have a content-sharing arrangement.