Autonomous vehicle (AV) firms have collected tens of billions of dollars to build genuinely self-driving automobiles. However, industry leaders and experts suggest that remote human supervisors may be required permanently to assist robot drivers in difficulty.
Much of the research and investment in autonomous cars have been motivated by the fundamental idea that computers and artificial intelligence will significantly minimize accidents brought on by human error.
However, there is a catch: It is tough to create robot vehicles that can drive more securely than humans since self-driving software systems simply do not have the human capacity for rapid risk assessment and prediction, especially in the face of unforeseen accidents or “edge cases.”
When asked if he could imagine a day where remote human overseers might be eliminated from operations, Kyle Vogt, CEO of Cruise, a division of General Motors, responded, “Well, my question would be, ‘Why?'”
Because there is always a live person available to assist, Vogt added, “I can give my consumers peace of mind.” I can’t think of a good reason to get rid of that.
The long-term requirement for remote human operators has never before been addressed by Cruise.
When posed the same question, Waymo and Argo, two products of Alphabet Inc. supported by Ford Motor Co. and Volkswagen AG, choose not to respond.
After a collision in San Francisco that wounded two people in June, GM recalled and upgraded the software on 80 Cruise self-driving cars this month. U.S. safety authorities said that the recalled software might “incorrectly forecast” the route of an approaching car, but Cruise insisted that the strange event wouldn’t happen again following the upgrade.
Some people’s skepticism regarding technology grows due to the possibility that human supervisors will continue to exist.
The optimistic implementation timelines for truly driverless vehicles were forecast just a few years ago, but they are already far behind.
For a completely autonomous vehicle without a steering wheel, brake, or accelerator pedal that would join its commercial ride-sharing fleet in 2019, General Motors applied for U.S. regulatory certification in 2018. The Cruise Origin that car is currently not expected to start production until spring 2023, Vogt said.
Elon Musk, the CEO of Tesla Inc., stated that his company would sell a million robotaxis “next year for sure” in 2019, even though its “Full Self Driving” vehicles have drawn criticism for not being able to drive independently without a human at the wheel and prepared to take manual control in an emergency.
Musk stated that creating self-driving vehicles was “far tougher than I originally expected, by far” in a YouTube interview in June. However, when pushed for a timeframe, he responded that Tesla may succeed “this year.”
A request for comment on this article from Tesla was not met with a response. The stakes for the AV business have increased due to the unfulfilled promise of total autonomy.
The CEO of Edge Case Research, which aids AV firms in risk assessment, management, and insurance, Mike Wagner, stated that if these businesses didn’t flourish during the following two years, they would cease to exist. At this point, “put up or shut up” is the only option.
Nowadays, many AV firms also employ safety drivers who sit in the driver’s seat and act as remote supervisors.
This remote personnel adds to the cost but aids in handling edge circumstances for self-driving automobiles. These might be as simple as an unexpected set of lane closures due to road work, or they could be more complex, like erratic, unpredictable conduct by pedestrians or human drivers.
Koosha Kaveh, CEO of Imperium Drive, which uses people as remote operators for automobiles in the English city of Milton Keynes, claimed that when a robot driver faces an edge situation, “it throws its hands up and says, “I don’t know what’s going on,” this personnel will eventually work as “air traffic controllers,” monitoring an increasing number of autonomous vehicles.
According to Vogt, fewer than 1% of the time, Cruise’s AVs operating in San Francisco now rely on people. But across hundreds, thousands, or even millions of AVs, that would equate to a sizable amount of time spent waiting for human direction while stationary on the road.
As more autonomous vehicles, which are more predictable than human drivers, crash, Imperium Drive’s Kaveh remarked – hit the roads, the number of edge cases will drop, “but you will never get to zero edge cases.”
Even decades ago, Kaveh continued, “fully autonomous vehicles won’t be 100% practical.”
But the level of competitiveness is increasing. Cities in China are seeking quicker approval of active AV testing.
The decline in investor investment in self-driving vehicles has also increased the need to address edge situations and reduce the price of everything from sensors to the number of humans involved in reaching the market.
Investors are uncertain when autonomous businesses can become lucrative, which has led to some skepticism. Trucks and last-mile delivery services using highways or predetermined, slow routes are likely to become profitable first, but it will still take years for them to do so.
According to investor website PitchBook, overall venture investment in future mobility businesses has decreased, with AV-focused companies being particularly severely hit, accounting for less than 10% of venture investment in the second quarter. (Graph: tmsnrt.rs/3Rzy04y)
In 2022, investment in mobility technology will decline. http://graphics.reuters.com/AUTOS-AUTONOMOUS/TECHNOLOGY/egvbkradkpq/chart.png
The quarter’s investment in AV companies decreased to $958 million. According to PitchBook, AV investment was flourishing only two years ago, with Alphabet’s Waymo raising $3 billion, Didi’s AV division raising $500 million, and Amazon.com Inc. acquiring AV company Zoox for $1.3 billion.
Rushing to the market
According to Chris Borroni-Bird, an independent consultant who oversaw advanced-vehicle initiatives at GM and Waymo, autonomous systems are less capable than people because their “perception and prediction algorithms are not as sophisticated as how a human brain thinks and chooses.”
For instance, when a ball rolls onto the road, which is harmless on its own, a person will believe it may be being followed by a child and will hit the brakes far more quickly than an AV, according to Borroni-Bird.
He continued, “I worry that AV firms will rush to market without demonstrating that safety is superior to human-driven cars.
Rushing to the market
According to Chris Borroni-Bird, an independent consultant who oversaw advanced-vehicle initiatives at GM and Waymo, autonomous systems are less capable than people because their “perception and prediction algorithms are not as sophisticated as how a human brain thinks and chooses.”
For instance, when a ball rolls onto the road, which is harmless on its own, a person will believe it may be being followed by a child and will hit the brakes far more quickly than an AV, according to Borroni-Bird.
He continued, “I worry that AV firms will rush to market without demonstrating that safety is superior to human-driven cars.
Driving-free delivery
AB Dynamics uses a robot arm to test vehicles on the track that it intends to add to slow-moving mining and agricultural trucks to make them mostly autonomous.
For example, Routh envisions a remote team of humans managing a fleet of self-driving mining vehicles that operate in enclosed spaces.
In quicker, more open areas, he does not see that scenario working for automobiles since it could be challenging for distant human supervisors to respond to risks promptly.
With support from remote human supervisors, British online food delivery and technology business Ocado Group Plc will introduce a small fleet of driverless delivery cars over the next 12 months. These vehicles will travel predetermined routes on a limited number of streets in a small UK city and never drive at speeds above 30 miles (48 km) per hour.
