Anthony Levandowski is aware of his reputation.
One of the most controversial figures in autonomous driving, his background has been well-covered, from building the Ghostrider motorcycle that competed in the DARPA Grand Challenge in 2004 and 2005, to his work on Google’s self-driving car.
It continues with his founding of Otto, a self-driving trucking startup acquired by Uber, to his role at the center of the Waymo vs. Uber case. Levandowski was accused of stealing Waymo’s self-driving secrets and bringing them along to Uber, which fired him. Waymo eventually accepted a settlement of $245 million.
There’s more of course, including the founding of a AI-focused religious organization, Way of the Future. It seems with Anthony Levandowski, there’s always more…
But Levandowski is only looking forward. He introduced his newest venture, Pronto, last year, and he’s ready to talk. Like Otto, Pronto focuses on self-driving in the trucking industry. But the outlook — and the tech — is different this time around.
Levandowski and Pronto grabbed headlines in December with a cross-country trip in a Toyota Prius that appears to set a record for the longest recorded journey of a totally autonomous car trip. After two other attempts that came to an end in Utah, Levandowski completed a 3,099-mile October trip from San Francisco to New York, only taking control of the vehicle to stop or refuel.
Electrek talked with Levandowski — along with Pronto.AI co-founder Ognen Stojanovski — about that cross-country trip, the goals of Pronto, and that reputation of his.
Levandowski claims that most people seem to be convinced of the veracity of his cross-country trip, though there have been critics. An article in The Guardian detailing the journey also included some voices expressing skepticism. Levandowski said he’s offered the chance to watch the entire video at normal speed (which would take more than 50 hours).
“Less than 5 percent of folks are like, ‘Is this really real? Did they really do it?’ I’m not really worried about that,” Levandowski said. He claims it would be easy to replicate. “We could do it again next week if we wanted to. It’s not a difficult-to-recreate event.”
Pronto’s tech is based around cameras and software — the company’s setup includes 360-degree video coverage of the outside of the vehicle, and another camera focused on the driver.
“We actually think that the driver monitoring piece is as important as the self-driving piece,” he said. “To make sure that the driver is paying attention and not using their cell phones….we look at where the head is, where are the arms pointing?”
Hence, the Copilot name of Pronto’s first system. “We think the driver is an integral part of how to make this current technology actually achieve the benefits safety-wise and fuel economy-wise that we think it can,” he said.
Levandowski compared the tech and its effect directly to the similar Tesla Autopilot system. If that sort of driving assistance helps on a short commute, he said, imagine how much of an impact it would have on a long drive.
What Pronto doesn’t use is Lidar or HD maps. Levandowski referred to these as “crutch technologies” in a Medium post announcing his return with Pronto. (Elon Musk has also referred to Lidar as “a crutch” in the past.)
“We think Lidar and the whole HD map craze that’s happened in the last five years or so is not going to be scalable for a couple reasons,” he said. “On the Lidar side, because of the cost, and the reliability. If you’re gonna have a Lidar, you’re going to need to go driverless. And going driverless is really far away.”
On this point, Levandowski is convinced — truly autonomous driving is still far off. He believes Pronto’s focus on Level 2 automation is the way to go, at least for now.
As for maps, Levandowski said they just aren’t quick enough to adjust to a changing world. He cited the time a self-driving Uber was filmed running a red light on its launch day. A crosswalk was unaccounted for, and the car went right on through.
“And that just wasn’t in the map, which is why the car drove through that,” he said. “If you look at it, maps will always have faults in them. And so, to make a true technology that you can deploy, you want to avoid doing that.”
Copilot Gears Up
Pronto will sell Copilot as a $5,000 aftermarket installation in Class 8 vehicles. Co-founder Stojanovski said Pronto does have some partners lined up. Those announcements are “a matter of months” away.
Levandowski mentions in his Medium post that Copilot is Pronto’s “first product.” Does that mean others are planned? He stays focused on the present.
“We think the passenger market is well served, for now,” he said. “We want to keep adding more and more functionality as time goes on to Copilot.”
Not included in those plans is removing the driver. “Version 1 will not allow you to drive from San Francisco to New York the way we demonstrated the technology could,” Levandowski said, but he noted that more features will be added over time.
Levandowski was quick to bring up safety at all points during our interview, and Pronto’s website is the same way. The first sentence you read on the site presents the company’s mission as “to deliver the safety benefits of self-driving technology for the benefit of all road users.” There’s also a webpage dedicated to safety.
In his Medium post, Levandowski wrote:
“One thing I continue to stand by is my safety record, which is second to none in this industry. My teams have a demonstrably impeccable track record. At every company where I have worked, safety was better during my tenure than before or after. On the safety front, I always look forward and am proud to work with and be challenged on a daily basis by the best in the business.”
But Levandowski’s public history with safety is shakier, much of that stemming from an October 2018 New Yorker article. The story quotes him: “If it is your job to advance technology, safety cannot be your No. 1 concern … If it is, you’ll never do anything. It’s always safer to leave the car in the driveway. You’ll never learn from a real mistake.”
The article goes on to recap an event in which Levandowski rides with another Google executive in a Prius, noting that “Levandowski had modified the cars’ software so that he could take them on otherwise forbidden routes.” Levandowski swerved away from a careening Toyota Camry, causing the other executive in the Prius “to injure his spine so severely that he eventually required multiple surgeries.”
The New Yorker article continues:
“The Prius regained control and turned a corner on the freeway, leaving the Camry behind. Levandowski and Taylor didn’t know how badly damaged the Camry was. They didn’t go back to check on the other driver or to see if anyone else had been hurt. Neither they nor other Google executives made inquiries with the authorities. The police were not informed that a self-driving algorithm had contributed to the accident.”
“I think if you look at the events reported in the New Yorker, they’re false,” Levandowski said, referring to the incident. He maintains it was blown out of proportion. “If you were to look at the video, you would wonder why it’s a big deal. There’s no swerving, there’s no contact, there’s no nothing. Basically, somebody tried to lane change and they realized there was a car there, and then…lost control.”
Levandowski asserts that he does indeed stand by his safety record. He claims a number of safety measures he put in place at other companies were later removed, which “caused bad things to happen after I was no longer involved.”
As for the New Yorker quote about safety not being of the utmost concern, Levandowski said it requires further explanation — that it’s always safer to not go out and try any testing in the real world. “The problem here is that the description of how to properly and safely develop technology is not suited to a quick sound bite,” he said.
Stojanovski pointed out that the tech and trucking industry are much different, and there’s no reason to push the envelope in a way that could lead them down an improper path. There’s no desire for any sort of “cool factor” in commercial trucking.
“That’s not happening with the kind of customers we talk to,” Stojanovski said. He notes that the trucking industry appreciates straightforward talk, and that Pronto has “already been accepted by the trucking industry, by and large.”
“To think that I don’t care about safety is just wrong,” Levandowski added.
Staying In The Game
Levandowski’s Medium post claims he’s “been painted into a villainous caricature,” words that he stands by.
“I definitely think that’s true if you look at the events and what people accuse me of…or even just the New Yorker article I think is a good example of that,” he said. “When you read that, you’re like, ‘Oh wow, this guy really doesn’t care about safety, look at that.’ There was a Twitter handle that somebody told me later on was calling me, like, ‘murderous asshole.’”
It’s unclear what Twitter account Levandowski was referring to, but the sentiment, he said, is not based on real facts.
“I understand how that came about,” he said. “I recognize I’ve been pushing for deploying technology out. And I’ve kind of rubbed people the wrong way in how to get technology into the market, I’ve learned a lot from there. But I’m actually a pretty nice guy. You’d imagine that I’m like this evil super villain or what somebody would call an evil mastermind genius or something like that. That’s not really a thing. I just care about the technology a lot and I want to get it out and I don’t really have a lot of patience for bureaucracy.”
He stresses that he remains in the self-driving game for the long haul. “(This) is something I love,” he said. “I think about this every day. I dream about it at night, I brush my teeth, I’m thinking about it. It’s kind of in my DNA.”
Though Levandowski did express some disappointment in how slow development has been in autonomous driving as a whole, he remains focused on his goals with Pronto.
“I’ve earned a reputation, rightfully or wrongfully,” he said. “At the end of the day I think haters are going to hate, but if you actually build something that is valuable and delivers safe improvements to an operation, they’re going to be really excited about using that. And that’s what we’re focused on.”