Looking forward to taking a self-driving taxi? Don’t hold your breath


  • TECH
  • Thursday, 26 Sep 2019

Waymo uses machine learning to help self-driving systems evaluate whether a dog is on a leash or whether the body language of a pedestrian suggests that they could step onto the road. — dpa

Self-driving cars will someday make having your own car, as well as much of the urban space devoted to parking, redundant.

Autonomous transportation is only five years away, experts keep saying. But arriving at the future is taking a bit longer than expected.

Just a few weeks ago, the US car manufacturer General Motors had to admit that it wouldn’t be able to begin its driverless taxi service this year after all. But it will expand testing with self-driving cars in San Francisco, said Dan Ammann, head of GM’s autonomous vehicle division, Cruise. The company didn't offer a new launch date.

At Volkswagen, autonomous vehicle manager Thomas Sedran admitted previous timelines had been too optimistic. "The robo-taxi will not significantly change the actual business model of the automotive industry until at least 2030," he told Germany’s Manager Magazin.

That will give the auto industry a bit more time to adjust their business models. It’s expected that the majority of profits from autonomous vehicle will generated by transport service providers, not by manufacturers.

Last year Cruise estimated that each autonomous vehicle could generate several hundred thousand dollars over its lifetime. That makes the sale amount of an individual car seem quite small.

It remains unclear how many vehicles robo-taxis will actually take off the roads. Analysts at investment bank UBS say with self-driving cars, New York one could get by with a third of today's taxi fleet. But that would also mean higher wear and tear, so a car might only last around three years. That need for replacement could actually boost production.

UBS analysts also dialled back their forecast for 2030 to 11 million autonomous vehicles on the roads worldwide from 26 million previously.

One major challenge is predicting the behaviour of other road users, says Drago Anguelov, head of research for Google’s Waymo. So Waymo uses machine learning to help self-driving systems evaluate whether a dog is on a leash or whether the body language of a pedestrian suggests that they could step onto the road.

The first deadly autonomous taxi accident, when an Uber test car in Arizona ran over a woman who was pushing a bicycle across the road in the evening, has further intensified the focus on safety.

Waymo is already testing autonomous taxis in Arizona without a safety driver at the wheel, but a human employee is always close by. There are still situations in which a computer reacts worse than a human. Waymo related a case in which a robot car showed up in front of a passenger who ordered it, but there was a cactus hedge between the person and the car.

Experts expect the technology to eventually arrive into everyday life. "I can imagine robot taxis on certain routes in the middle of the next decade," says Axel Schmidt of Accenture – five years again. The stakes are high: "The first person to break through this problem and offer a robotic taxi will have a clear market advantage and a massive influence on the industry.” – dpa


Article type: metered
User Type: anonymous web
User Status:
Campaign ID: 7
Cxense type: free
User access status: 3
   

Across The Star Online