Otto’s Autonomous Trucks Drive Almost 24/7 to Refine Technology of Tomorrow
An overladen pickup abruptly veered in front of our big-rig cab as we headed south on Interstate 280 from San Francisco. Luckily, we slowed just in time.
Sitting in the back of the truck’s cab, Eric Berdinis drew a breath. “Whew, we got past that fine,” he said.
Our truck, a Volvo VNL 780 tractor cab, was driving itself.
Tricked out with dozens of sensors, the truck constantly was detecting the presence of other vehicles; the lanes, curves and exits on the road; and its own motion. It used cameras, motion sensors, GPS, radar and lidar, a version of radar that uses laser light instead of radio waves, plus sophisticated software to do the job a human driver would use eyes and ears for.
As legally required, a licensed commercial Class A driver, Nic Munley, sat in the driver’s seat, hands hovering over the steering wheel, ready to take over at any time. In the co-pilot seat, Brian Gagliardi monitored a laptop display of what the truck “saw” as it barreled down the highway in self-driving mode.
The self-driving technology was devised by Otto, a San Francisco startup that was acquired by Uber this month for a reported $680 million and is now an integral part of Uber CEO Travis Kalanick’s race into a future where taxis, delivery vehicles and longhaul trucks can pilot themselves. That race, with deep-pocketed contestants including Google, Apple, Tesla, Ford Motor Co., General Motors Co., Lyft, Delphi and other carmakers and technology companies, is accelerating with Uber’s plans to offer robot taxi rides with backup drivers in Pittsburgh this month.
Otto’s founders, two of whom previously worked on Google’s self-driving car project, started the company in January and had a self-driving truck on the road in March. They believe their approach, a kit to retrofit trucks, is faster and cheaper than designing and making a vehicle from scratch. They see commercial trucking, which hauls about 70% of U.S. freight over 220,000 miles of highways, as a crucial and manageable problem to tackle.
“Highways are simpler by nature, with less complexity to deal with: no pedestrians, no crosswalks, no traffic lights,” co-founder Lior Ron said Aug. 25 shortly after returning from Pittsburgh, where he and other Otto team members met with Uber’s autonomous driving team. “Commercial transportation is the backbone of the U.S. economy but hasn’t benefited much from Silicon Valley technology advances. It faces many pain points and challenges. We can solve that, deploy and bring the solution to market sooner rather than later.”
Ron said the system would be ready for prime time in much less than five years. In the meantime, Otto CEO Anthony Levandowski will lead Uber’s self-driving efforts in Pittsburgh, San Francisco and Palo Alto, Calif.
For the foreseeable future, Otto says there will be a driver in the truck to handle everything that happens off the highway: entering and exiting, refueling, loading and unloading freight, interacting with police at weigh stations.
But that driver could take a nap in the back — or grab a full night’s rest — while the truck pilots itself down the highway, resulting in much better utilization of vehicles that no longer have to sit idle while drivers rest. “A driver can go to sleep in El Paso and wake up in Dallas nine hours later, fresh and ready to keep driving,” said Berdinis, product lead at Otto.
Otto’s target customers are independent truckers who drive their own vehicles. About two-thirds of the nation’s 500,000 trucking companies are run by owner-operators — often a single person with a longhaul big rig that may have cost around $100,000.
“The technology is almost like a superpower for them,” Berdinis said. “They have the most to gain with something like this. They get more hours on the road. They’re safer and more fuel-efficient.”
But allowing the driver to leave the wheel still faces regulatory and technical challenges, exacerbated by May’s fatal crash of a Tesla in Autopilot mode, where neither the driver nor the system detected the white side of a tractor-trailer against a brightly lit sky.
For now, Otto is focused on logging as many test miles as possible with its five (soon to be six) retrofitted Volvo cabs that it runs almost 24/7 on highways in the Bay Area and elsewhere. (Drivers keep control on city streets.) It has trailers for testing but often rolls out just the cabs, as they are easier to maneuver. It soon will partner with some fleets to retrofit them for more testing, Ron said.
Otto’s 100-plus person team works out of a sprawling 50,000-square-foot garage, work space and office within a couple of blocks of both Highway 101 and I-280. The space still has a sign from its former use as an auto repair shop and showroom a half century ago. Otto not only designs but produces many of the components for the kits in an on-site full machine shop. “If an engineer has an idea, we can machine a prototype that day, as opposed to waiting a few weeks,” Ron said.
Autonomous driving uses similar sensors for trucks as for cars, But there’s a key difference in how those sensors control the vehicles. With cars, engineers can tap into electronic control systems to send signals to turn left or right, for instance. With trucks, the controls are manual. Otto had to add its own air pressure lines to control braking, for instance. There are some advantages, though: For example, big-rig trucks’ height allows sensors to perceive road conditions far ahead. “We’re kings of the road with high vantage points,” Ron said.
Unlike Tesla’s system, which relies on constant communication with the internet, Otto’s technology is self-contained beyond picking up satellite GPS signals. “The truck is just as capable of driving itself in an area with no [internet] coverage as in an area with great coverage,” Berdinis said.
Once there are fleets of autonomous vehicles, they may communicate with one another. For instance, once a car figures out how to navigate around a construction zone on the road, it could “teach” all the trucks behind it how to do so.
Ron, who previously was product lead for Google Maps, said mapping is a key issue. The trucks always will have an underlying map but will be able to update it dynamically as road conditions change.
The biggest technical challenge, Berdinis said, is the other drivers on the road. “You can never predict what they’re going to do. A lot of people realize at the last second that they want to change lanes, and cut you off. Crazy things happen on the road, so we need to rack up more miles. The more we drive, the more we see.”
Otto’s first truck took six weeks to retrofit, a period it’s whittled down to six days. The vision is that an owner-operator would drive his or her truck into a station and pick it up later that day, ready for autonomy. While costs are unknown, Otto might offer a per-mile subscription model to make it more affordable.
Otto’s partnership with Uber will go beyond joining forces on autonomous technology. It also will leverage Uber technology to improve how trucks and cargo get connected. “Drivers will be able to operate Uber-style and always find the next shipment; shippers will be able to press a button and get a truck,” Ron said.
Sandeep Kar, global vice president of research at Frost & Sullivan, said he sees the Otto-Uber partnership as a win-win. “It solves two problems,” he said. “One is autonomous trucks at mass scale and the other is creating an ‘Uber for trucks’ to increase efficiency.”
Otto and Uber have a lot of obstacles to dodge on the road ahead. But with autonomous vehicles looking inevitable, they’ll just keep on trucking.
