MILTON KEYNES, England, September 12 (Reuters) – Autonomous vehicle (AV) startups have raised tens of billions of dollars on promises to develop truly self-driving cars, but industry executives and experts say that remote human supervisors might be permanently needed to help troubled robot operators.
The central premise of autonomous vehicles – that computers and artificial intelligence will dramatically reduce accidents caused by human error – has driven much of the research and investment.
But there’s a catch: It’s extremely difficult to make robotic cars that can drive more safely than humans, because autonomous software systems simply lack the ability of humans to predict and quickly assess risk, particularly in case of unexpected incidents or “extreme cases”.
Join now for FREE unlimited access to Reuters.com
“Well, my question would be, ‘Why? ‘” Kyle Vogt, CEO of Cruise, a unit of General Motors (GM.N), said when asked if he could see a point where remote human monitors should be removed from operations.
“I can offer my clients peace of mind knowing that there is always a human to help them if needed,” Vogt said. “I don’t know why I would want to get rid of it.”
This is the first time Cruise has acknowledged the long-term need for remote human operators.
Like air traffic controllers, these human supervisors can be seated tens of hundreds of miles away monitoring the video feeds of multiple AVs, sometimes with a steering wheel, ready to step in and get stuck robot drivers to move again – the AVs. invariably stop when they can’t figure out what to do.
Alphabet Inc’s Waymo and Argo (GOOGL.O), which is backed by Ford Motor Co (FN) and Volkswagen AG (VOWG_p.DE), declined to comment when asked the same question.
GM recalled and updated software for 80 Cruise self-driving vehicles this month after a June crash in San Francisco left two people injured. U.S. safety regulators said the recalled software could “incorrectly predict” the path of an oncoming vehicle, and Cruise said the unusual scenario won’t happen again after the update. Read more
For some, the idea that human supervisors might be here to stay raises more doubts about the technology.
Truly autonomous vehicles are far behind the optimistic rollout schedules predicted just a few years ago.
In 2018, GM sought US government approval for a fully self-driving car without a steering wheel, brake or gas pedal that would enter its commercial ride-sharing fleet in 2019. This vehicle, the Cruise Origin, is not expected to begin production before spring. 2023, Vogt said.
In 2019, Tesla Inc (TSLA.O) CEO Elon Musk promised a million robotaxis “next year for sure” – although his company’s “Full Self Driving” offer was criticized because that its cars are not capable of driving themselves without a human behind the wheel and ready to take manual control in an emergency.
In a YouTube interview in June, Musk said developing self-driving cars was “far more difficult than I initially thought”. But when asked for a timeline, he said Tesla might do it “this year.”
Tesla did not respond to a request for comment for this story.
The unfulfilled promise of true autonomy has raised the stakes for the audiovisual industry.
“If these companies don’t succeed within the next two years, they won’t exist,” said Mike Wagner, CEO of Edge Case Research, which helps audiovisual companies assess, manage and insure risk. “It’s a case of put up or shut up at this point.”
REMOTE HUMANS WATCHING
Many AV startups today use humans as remote supervisors, alongside safety drivers sitting behind the wheel. Read more
These remote humans are an added expense, but help self-driving cars handle extreme cases. These could include something as basic as an unknown set of lane closures during road construction, or erratic and unpredictable behavior by pedestrians or human drivers.
When a robot driver encounters a fringe case, “he raises his hand and says, ‘I don’t know what’s going on,'” said Koosha Kaveh, CEO of Imperium Drive, which uses humans as remote operators. for cars in the English city of Milton Keynes. Over time, these people will act as “air traffic controllers”, overseeing an increasing number of self-driving cars.
Cruise’s Vogt says the company’s AV vehicles on San Francisco roads currently rely on humans less than 1% of the time. But on hundreds, thousands, or even millions of AVs, that would be a significant amount of downtime on the road waiting for human guidance.
Imperium Drive’s Kaveh said that as more self-driving cars – which are more predictable than humans – hit the road, the number of extreme cases will decrease, “but you’ll never get to zero cases.” .
“Even decades from now, you won’t get truly 100% autonomous vehicles,” Kaveh added.
Nevertheless, the competition is increasing. Some Chinese cities are pushing to allow active AV testing faster.
The need to tackle edge cases and reduce the costs of everything from sensors to the number of humans in the loop in order to get to market has also intensified because investor funding for self-driving cars has increased. fall.
Doubt has crept in as investors wonder how quickly standalone businesses will become profitable. Simpler or slower AV vehicles, such as trucks or last-mile delivery services operating on highways or low-speed routes, are likely to reach profitability first, but will still take years to get there. to arrive at.
Overall investment in future mobility startups has slowed, with audiovisual-focused companies particularly hard hit, accounting for less than 10% of venture capital investments in the second quarter, according to investor website PitchBook. (Graphic: https://tmsnrt.rs/3Rzy04y)
Investments in audiovisual startups in the quarter fell to $958 million. Just two years ago, AV investments were booming, as Alphabet’s Waymo raised $3 billion, Didi’s AV unit raised $500 million and Amazon.com Inc (AMZN.O ) has acquired AV startup Zoox for $1.3 billion, according to PitchBook.
Autonomous systems aren’t as capable as people because their “perception and prediction algorithms aren’t as good as how a human brain processes and decides,” said Chris Borroni-Bird, an independent consultant who ran previously advanced vehicle programs at GM and Waymo.
For example, a human seeing a ball rolling down the road — harmless in itself — will assume it might be followed by a child and brake much faster than an AV, Borroni-Bird said.
“I worry that AV companies will rush into the market without proving that safety is better than human-driven vehicles,” he added.
The problem is that there are “tens of billions of potential edge cases” AVs could encounter, said James Routh, CEO of AB Dynamics
Automotive data startup Wejo Group Ltd (WEJO.O) receives 18 billion data points daily from millions of connected cars and contributes to simulations for AV vehicles, said Sarah Larner, executive vice president for strategy and innovation.
“But there are so many variables like weather, you can take a borderline case and then have to layer all the different variations,” she said. “It’s really millions of releases.”
In its on-track tests for cars, AB Dynamics is using a robotic arm that it plans to retrofit on slow-moving mining and agricultural trucks to make them largely autonomous.
Routh envisions a remote team of humans overseeing fleets of, say, self-driving mining trucks operating in closed environments.
He doesn’t see this scenario working for vehicles in faster, more open environments, as it could be difficult for remote human supervisors to react to hazards quickly enough.
Over the next 12 months, UK food delivery and online technology company Ocado Group Plc (OCDO.L) will roll out a small fleet of driverless delivery vehicles with autonomous vehicle software start-up Oxbotica – backed by remote human supervisors – who will operate on just a few streets on set routes in a small UK town and never drive at speeds in excess of 30 miles (48 km) per hour.
“At 30 miles an hour, if a vehicle panics, it can hit the emergency brake and call for help,” said Alex Harvey, head of advanced technology at Ocado. “It looks like a very viable low-speed strategy.”
“But you can’t play this game on a freeway,” Harvey added, because sudden stops in extreme cases would pose a safety risk.
Harvey said it would take about five years for Ocado to develop a cost-effective driverless delivery system. More than half of Ocado’s UK customers could be reached with AVs at speeds of no more than 40mph, he said. Eventually, the service could be rolled out to Ocado customers such as US retail chain Kroger Co. (KR.N)
Join now for FREE unlimited access to Reuters.com
Reporting by Nick Carey in Milton Keynes, England, and Paul Lienert in Detroit Editing by Ben Klayman, Matthew Lewis and Louise Heavens
Our standards: The Thomson Reuters Trust Principles.
Truly self-driving cars may be impossible without helpful human contact