VOA新闻杂志2022--自动驾驶汽车可能需要人工帮助（在线收听）

Self-Driving Cars May Require Human Help

Makers of self driving, or autonomous, vehicles have raised tens of billions of dollars based on promises to develop the fully robotic product. However, industry leaders and experts say the technology may forever require human supervision.

Supporters of autonomous vehicles, or AVs, say that computers and robotic technology will reduce the number of traffic accidents. But in reality, making self-driving cars safer than human-operated is complex. Self-driving programming lacks the human ability to predict and recognize risk quickly.

Kyle Vogt is the head of Cruise, a unit of American car company General Motors (GM). When asked if he believed humans could ever be completely free of vehicle operation, Vogt questioned the worthiness of such a goal.

"I can provide my customers peace of mind knowing there is always a human there to help if needed," he said. "I don't know why I'd ever want to get rid of that."

GM recalled and updated software in 80 Cruise self-driving vehicles this month after a crash in June. Two people were injured in the accident in San Francisco, California. U.S. safety officials said the recalled software could "incorrectly predict" an oncoming vehicle's path. Cruise said its vehicles would not make the same mistake again after the update.

For some, the need for human supervision increases doubt about the technology. And entirely self-driving vehicles are far behind in development that industry leaders have promised.

In 2018, GM sought U.S. government approval for a fully autonomous car. It had no steering wheel or brake or gas pedals. It was to be marketed in 2019. But that vehicle, the Cruise Origin, now is not expected to begin production until spring 2023, Vogt said.

In 2019, Tesla head Elon Musk promised 1 million robotaxis would be in place by 2020. His company's "Full Self Driving" feature has been criticized because its cars use human operators.

In June, Musk said that building self-driving cars had been far more difficult than he had expected.

Mike Wagner is with Edge Case Research, which helps AV companies analyze risk.

He said: "If these companies don't succeed over the next two years, they're not going to exist anymore."

A human eye is watching

Many AV companies today use humans as remote piloting supervisors. They support self-driving cars in dealing with unexpected events on the road. The industry calls these "edge cases."

Edge cases could include street closures for roadwork, or unpredictable actions by a human driver or walker.

When a self-driving car experiences an edge case, "it puts its hands up and says, 'I don't know what's going on,'" said Koosha Kaveh. He is with Imperium Drive, which is using humans as remote operators for cars in the British city of Milton Keynes. Kaveh said their work is similar to air traffic controllers but for autonomous cars instead of planes.

Cruise's Vogt says the company's AVs on the roads in San Francisco currently depend on humans less than one percent of the time. But across thousands or even millions of AVs, that would add up to a large amount of time stopped on the road waiting for human guidance.

‘Rush to market'

Autonomous systems are not as effective as people because their "perception and prediction algorithms are not as good as how a human brain processes and decides," said expert Chris Borroni-Bird. He has worked with self-driving vehicles with GM and Waymo.

For example, Borroni-Bird said a human seeing a ball roll into a road is likely to immediately recognize that a child might be chasing that ball. AVs will not. So, a human driver will reduce speed far more quickly than an AV in such an edge case.

That worries Borroni-Bird.

"I am concerned that AV companies will rush to market without proving the safety is better than human-driven vehicles," he said.

Words in This Story

get rid of — v. to do something so that you no longer have or are affected or bothered by

recall — n. an official order for someone or something to return

pedal — n. a flat piece of metal, rubber, etc., that you push with your foot to make a machine move, work, or stop

remote — adj. connected to a computer system from another place

perception — n. the way you think about or understand someone or something

algorithm — n. a set of steps that are followed in order to solve a mathematical problem or to complete a computer process