CONCORD, CALIF. — The self-driving semi-truck will be just like the self-driving car ... except not really.
For starters, the automated truck might be around sooner. Though a litany of tech and auto giants are racing to put self-driving taxis, passenger cars and shuttles on the streets, those projects face some hurdles in the form of regulatory approval, and hardware and software development.
But Otto, a company led by former Google employees focused solely on self-driving trucks, thinks it’s found a way around all that. Federal safety standards require self-driving vehicles to have brakes and steering wheels? Otto’s truck has those. California regulators want automated vehicles to only operate under government-approved conditions? For now, Otto is only looking at using automated, or AV, mode on highways. Automakers need to redesign vehicles that incorporate the necessary hardware to allow the car to drive itself? Otto’s current technology is aftermarket — it simply wants to outfit existing trucks and send them out onto the road.
“This shouldn’t be a 10-year exercise — probably not even a five-year exercise,” Otto co-founder Lior Ron said at a press event Aug. 8. “We believe the basic building blocks of the technology is real. We need to thoughtfully integrate them and bring the solution to the market. The aftermarket approach, the kit approach, allows us to start deploying that on existing trucks versus waiting for a new cycle of trucks to come from the factory. So we believe we can start deploying the technology commercially in the next couple years.”
Of course, the company still has some distance to go. At the moment, it’s just beginning to try out what Ron calls “edge cases” — circumstances that are better tested in a controlled environment first — at the self-contained GoMentum Station testing grounds near the San Francisco Bay Area town of Concord, Calif. The station has already opened itself up to other self-driving car manufacturers like Honda; as of last week, it’s playing host to big rigs as well.
“Our trucks are like kids, and they’re taking baby steps now," Ron said, "and GoMentum is the perfect playground for them to take those steps safely."
As it prepares the technology for deployment, Ron and fellow co-founder Anthony Levandowski said they’ll be looking to connect with transportation officials across the country to work out how the technology will impact them. Here are four ways the duo thinks self-driving trucks will improve transportation.
As with nearly everybody involved in self-driving cars, the first thing Levandowski and Ron turn to when discussing their product is traffic fatalities. Since research has repeatedly shown that almost all crashes happen because of mistakes humans make while driving, industry leaders are betting that computers can react faster and smarter and avoid many — if not most — of those collisions.
Otto's self-driving truck, showing sensors that would be installed high atop existing trucks that offer vehicles an unobstructed view of the road ahead. Photo by Ben Miller.
But semi-trucks are unique in the world of transportation. They’re large, so they take a longer time to decelerate than smaller vehicles. They have big blind spots when carrying trailers. The drivers spend lots of time on the road, pitting the need for sleep against the need to deliver on time.
With trucks driving themselves, humans can catch up on sleep and avoid taking the wheel while they’re tired. And since self-driving trucks will necessarily be equipped with a suite of sensors, the software should actually have better visibility than a human driver.
“You have a better point of view. You have different sensor modalities,” Levandowski said. “So, humans are just doing optical. We have this amazing brain that can extract all of this information from the data. But if everybody had the point of view where those cameras are, you could see a lot [more.]”
Self-driving trucks will not only be automated, they’ll be connected too. The sensors that take measures of the environment will most likely be capable of connecting with other vehicles and infrastructure, which means they will be able to share data that can help people make decisions about traffic both immediately and in the long run.
It's an idea that the Colorado Department of Transportation will begin testing out soon on one stretch of highway running through mountain passes: By sharing information between vehicles about what’s ahead, cars might be able to avoid traffic or other kinds of slowdowns.
In the future, government might be able to tap into some of that data as well — whether that means looking at trends that can inform planning decisions or simply being aware of hazards in the road.
“The more self-driving vehicles are on the road, the more information they can share with each other in real time,” Levandowski said. “So you have a little more of a good idea of where traffic is going on, like, ‘Hey, lane two is slow and now you should be driving in lane three.’ Or realizing there’s construction up ahead, there’s a merge, there’s been an accident, there’s a car stalled on the highway. These are all things that can be pulled out of sensors automatically as you’re driving by. So think of it as just like Waze, [which] allows people to share information human-wise. This is a way to make that happen automatically.”
In Ron’s eyes, traffic congestion isn’t just an annoyance — it’s an economic ball and chain.
Transportation experts tend to agree. An annual report from Texas A&M University describes congestion as a “tax” on commuters — every minute they spend in the car, inching forward and then stopping again, is another minute the driver didn’t spend working, or making dinner, or relaxing with family. Alongside the extra fuel spent idling, the money wasted on missed flights, rescheduled meetings and the like, the report puts the cost of congestion at $160 billion per year in the U.S.
Though trucks make up far less than half of the traffic on the road, Ron sees an opportunity for self-driving big rigs to help mitigate the problem by staying out of traffic.
Essentially it’s an efficiency proposition — it’s more efficient to drive trucks when traffic is flowing smoothly than when it’s backed up.
“Now the truck driver time doesn’t become a factor in scheduling those routes because the truck driver can actually sleep in the back, so you can start driving during nighttime instead of during peak commute hours,” he said. “You can start basically load-balancing the congestion and the traffic on the road.”
Because Otto is looking first at installing self-driving technology in existing trucks, it won’t necessarily lead to an increasing number of clean-running trucks. What it can do, though, is allow trucks to drive more efficiently.
Another company that makes frequent appearances at self-driving auto industry events is looking into a similar proposal — Peloton. The idea with that technology is to have trucks “platoon,” or follow very closely together so that the drafting effect reduces fuel use. To do that, the rear truck needs to have a certain level of automated driving ability.
Ron thinks platooning isn’t the only way to cut down fuel consumption in trucks, though. When a computer is in charge, he said, it can avoid a lot of the slowing down and speeding up that consumes fuel unnecessarily.
“The computer is basically determining the best driving conditions on that specific corridor,” he said. “It can take into account grade level, it can take into account traffic, it can produce basically a much smoother ride on the road. It can get you to the destination as fast as driving like 7, 10 miles per hour higher, which means much better fuel efficiency per mile.”