While it’s still unclear how soon self-driving truck technology will be road ready, how much it will cost or when the regulatory framework will be in place, you can bet Anthony Levandowski will be at the forefront of any discussions about the future of autonomous cars and trucks.
Google self-driving car veteran Levandowski is the co-founder of autonomous truck start-up Otto, which he set up earlier this year with Lior Ron, who previously headed Google Maps.
The San Francisco company is engineering kits to help retrofit big rigs use autonomous driving features. Ride-hailing company Uber Technologies acquired Otto in August and tapped Levandowski to lead its self-driving truck and car division. He will be at the forefront of Uber’s efforts to automate trucking and develop robotic taxis.
Levandowski, 35, has long been into robots.
When he was an engineering student at UC Berkeley, he won a 2001 technology contest by using 300 Lego pieces to build the BillSortBot, a bug-eyed simple robot that can sort Monopoly money, according to the university. A few years later, he built an autonomous motorcycle, nicknamed Ghostrider, for the Defense Advanced Research Projects Agency Grand Challenge. It was the only two-wheeled entrant in the autonomous vehicle races of 2004 and 2005 sponsored by DARPA, according to the National Museum of American History, which has Ghostrider in its collection.
Levandowski, who has a bachelor’s degree as well as master’s degrees in engineering and business from UC Berkeley, said he sees trucking as “the circulatory system of the American economy.”
Nearly 70 percent of all the freight tonnage moved in the U.S. goes on trucks, according to the American Trucking Associations. Trucking industry revenue hit a record $726.4 billion in the U.S. last year.
“Everything you see in this room came on a truck,” Levandowski said this week during a presentation on self-driving trucks at the American Trucking Associations’ Management Conference & Exhibition in Las Vegas.
Trucking, he said, is ripe for automation. Shipping goods could eventually work like Uber’s ride-sharing app, where “you press a button and a car shows up.” Shippers would do the same to arrange hauling freight.
“Liquidity is the ability to match carriers with shippers,” Levandowski said. “If you are delivering people, why not deliver food, why not packages? And if you are delivering packages, why not freight?” he said.
The trick is to ensure there are self-driving trucks everywhere its customers need freight.
For now, Uber is starting with people rather than freight. In September, it launched a pilot program to offer rides in 14 self-driving cars in Pittsburgh. The vehicles still have drivers behind the wheel in case they need to take control of the car, but the idea is to learn how to make a robotic taxi work.
“We are pushing for the technology to grow and be better than the best drivers,” Levandowski said. “Once you have a vehicle that can drive better than the best driver, the same thing as Uber, you want to be able to take that driver and put them on everything.”
The company’s system uses a variety of sensors including radar, lidar — a form of radar only with lasers — and cameras to paint a digital picture of the truck’s surroundings. Software will then determine speed, braking, acceleration and other functions as the truck moves down the highway.
Otto already has a fleet of six test trucks operating and plans to have more ready to hit the highways this year.
Analysts believe trucking and commercial transport could more quickly adopt self-driving vehicles than the consumer market.
Unlike an Uber driver or taxi, big rigs rarely have to navigate suburban streets. Trucks travel on highways, usually from factory to distribution center or warehouse to store, said Friedmar Rumpel, a director in the automotive practice of global consulting firm AlixPartners. That style of driving is more predictable and better suited to autonomous driving technology, he said.
Although Levandowski didn’t give an estimate of how much the technology is going to cost, he said it would “get cheaper over time.”
Uber may offer Otto technology at a significant discount for companies interested in utilizing Uber Freight, he said. The service will be launched in next year.
Otto and Levandowski have not provided many details. Otto has said on its website that it “is rethinking transportation with self-driving trucks and a new way to connect drivers with shippers.” It is currently signing up truck drivers and fleets to be part of its “early-access program” at testdrive.ot.to.
“It’s 2016, we now make phone calls that are automatically connected,” Levandowski said. “Why aren’t shipments automatically connected from the shipper to the carrier and vice versa?”
Levandowski said self-driving trucks reduce crashes. Truck drivers are more prone to make an error the more hours they drive, he said, citing that one in four drivers will have a “safety-critical” event when driving in the eleventh hour. Autonomous systems can relieve some of the driving burden, leading to less driver fatigue, according to Levandowski.
Federal regulators issued a new autonomous vehicle policy in September. When they introduced the new guidelines, traffic safety officials said that self-driving vehicles would be involved in fewer crashes than human-driven vehicles.
About 35,000 people died in roadway accidents in 2015, and 94 percent of the crashes “can be tied to a human choice or error,” according to the U.S. Department of Transportation, which jointly released the new guidelines with the National Highway Traffic Safety Administration.
Levandowski said he foresees a future in which autonomous trucks are built without cabs, so truck purchasers will get “30 to 40 percent more truck” because they won’t need to include truck drivers’ creature comforts like air conditioning, sleeper windshields, seat belts or heating.
Self-driving trucks will also help the industry address a chronic shortage of drivers, he said. Autonomous trucks can operate nearly 24 hours a day, ending worries about driver fatigue or violating federal hours-of-service rules.
ATA members attending the event expressed concerns about who would shoulder the liability in a crash involving a truck using self-driving technology or with a group of platooning trucks. Levandowski said electronic data collection systems on the trucks will help address the issue “on a case-by-case basis through litigation.”
The data would sort out who was to blame.
“If the vehicle catches on fire, maybe it was the vehicle manufacturer at fault,” he said. “If it’s a technology malfunction, maybe the technology manufacturer is at fault. If the driver is doing something they shouldn’t be doing, maybe the driver is at fault.”