Are self-driving cars so five minutes ago?
That would seem to be the takeaway from several recent headlines about the still-in-the-testing-phase technology. "Why self-driving cars could be going the way of the jetpack," read one headline in New Scientist magazine.
The suggestion here is that it's a possible technology and exists in some form. Still, because of difficulties, it has minimal commercial application.
A big reason for this concern is the investigation launched in August by the National Highway Traffic Safety Administration into the limited autopilot function of Tesla vehicles, which covers over 750,000 Teslas made since 2014.
Tesla Autopilot seems to have a problem with stopped emergency vehicles, and at least one case has allowed a car to plow right into one that was rendering assistance.
Democratic Sens. Richard Blumenthal of Connecticut and Ed Markey of Massachusetts sent a letter to the NHSTA calling for a "thorough investigation" in April after a fatal crash in Texas.
In August, the two members of the Senate Commerce, Science, and Transportation Committee released another joint statement saying that the agency's investigation "should inform the agency's recommendations on fixes the company must implement to improve the safety of its automated driving and driver assistance technology and prevent future crashes."
Tesla has said the autopilot function is not meant for self-driving. The NHSTA affirmed that around the time it commenced its investigation. "No commercially available motor vehicles today are capable of driving themselves. Every available vehicle requires a human driver to be in control at all times," a spokesperson for the agency told the BBC.
Another reason for the concern is that those in charge of developing the technology openly admit that they're hitting some roadblocks to a broader rollout. For example, the Google self-driving car in development, which is the furthest along in terms of fully autonomous passenger vehicles, is the Waymo.
You cannot buy Waymo vehicles, but you can hail a driverless taxi in Phoenix, Arizona, if you have the right app. They will mostly get you to where you're going but occasionally encounter unexpected hang-ups.
For instance, regular Waymo rider JJ Ricks takes trips in the Phoenix area, films his travels, and uploads them to YouTube. A video posted in May showed the Waymo taxi getting stumped by traffic cones and blocking traffic for 15 minutes. Finally, a Waymo support person showed up and got out of the jam by switching the vehicle to manual.
The slow pace of innovation has led to the departure of Waymo's CEO, chief financial officer, and chief development officer this year.
However, a pair of policy analysts told the Washington Examiner that these kinds of setbacks do not mean that self-driving cars are, in any sense, over.
Jeff Davis, a senior fellow at the Eno Center for Transportation, said there's both a technology problem and a problem of scale.
"Computerized AI is still a long ways away from being able to see a moving object and determine, within a quarter of a second, if that object is a human baby or an animal or a plastic shopping bag blowing in the wind or some other kind of harmless thing," Davis said. "The human brain can do that easily and does so all the time while driving, but AI is still a long way from being able to do near-instantaneous object recognition in context and thus being able to make the decisions about how to steer the car."
There is a way to fix that, but there's also a policy problem.
"The way to get better is to do more experimenting, which means more test vehicles on the road under carefully monitored conditions," Davis told the Washington Examiner.
The policy problem is that the trial lawyer lobby, the American Association for Justice, had blessed language in a 2018 bill to make that possible but then reversed course, killed the bill, and created a yearslong legislative traffic jam.
Marc Scribner, a transportation policy analyst for the Reason Foundation, chalked talk of self-driving cars being "over" to the fact that "a lot of the public discussion of transportation technology is dominated by people who have backgrounds in the Silicon Valley technology world or tech policy world, not in transportation or transportation policy, so they tend to adopt unrealistic innovation time frames that might be realistic for the latest 'killer app' but not for transportation."
Scribner believes that there's "still plenty of reason to be optimistic about the long-term prospects of [self-driving] technology, even though the initial hype bubble has burst. There's also work that policymakers can begin today so they are adequately prepared for deployment in the future."
Washington Examiner Videos
Original Author: Jeremy Lott
Original Location: Is the self-driving car doomed already?