Friday, October 14, 2022

Driverless Technology as of Fall 2022: Bleak, but Redirected

It’s been five years since the world was expecting an end, or nearly so, to people driving “meatmobiles,” a briefly-used name for human-controlled vehicles.  Since then, for various reasons from legislative balking to overreactions about driverless deaths, it has not seemed that way at all.  However, there are still useful pieces being written.  What has come out over the past four months?

The first was “Self-Driving and Driver-Assist Technology Linked to hundreds of Crashes, U.S. Data Shows,” by Neil E. Boudette and Cate Metz in the June 15th New York Times.  This one dealt mostly with the widespread spinoff from fully autonomous cars, “advanced driver-assistance technology” in conventional ones.  Per the authors, “over the course of 10 months” there were 392 National Highway Traffic Safety Administration “cataloged” accidents, with six people dead and five “seriously injured.”  Almost 70% of them involved Teslas and three-fourths of the rest were with Hondas.  A good point from a former Department of Transportation officer was that we don’t have a “baseline” for this information, so don’t really know if it’s high due to developing technology, excessive in general, or a reasonable tradeoff. 

Is it meaningful to say that “Driverless cars shouldn’t be a race” (Shira Ovide, The New York Times, August 11th)?  Apparently, companies and their observers are using that as a metaphor, for this wide-open and once highly promising area which requires standardization and benefits massively from cooperative seller attitudes and pooled efforts.  It’s especially damaging when undue fear of autonomous vehicles is a problem anyway, and the image of a road race without drivers is jarring.  So let’s go with “building the future,” or some such, instead.

There have been a few driverless mini-rollouts, and one was “Lyft Unveils Self-Driving Car Service in Las Vegas (With Caveats)”, by Cade Metz, once again in the Times.  It’s in a limited sort of structured area, that city’s Strip, which in its most developed parts has a relatively small set of turnoffs, has a minimal number of pedestrians on the road itself, and for various reasons is not as chaotic as it once was.  But the “caveat” that “reporters are not allowed to use these services without a driver behind the wheel” says something about Lyft’s confidence, and it isn’t good.  We’ll see how long this one lasts.

“As Driverless Cars Falter, Are ‘Driver Assistance’ Systems in Closer Reach” (Lawrence Ulrich, The New York Times, September 16th)?  We already have them, but here’s some mention of where this technology might be headed, that instead of moving toward being “entirely driverless,” it might become “more like a no-nonsense chaperone,” including monitoring drivers’ eyes and alerting them if they “look away for more than a few seconds.”  That may be the right way to go now, and perhaps later in this decade, when enough have unlearned the idea that partially autonomous systems can allow them to stop driving, emphasis can be returned to work-saving functionality.

One of the largest players in driverless technology late last decade is still around, but, per the content if not the title of “Chipmaker Nvidia launches new system for autonomous driving” (Reuters, as published in Fox Business, September 20th), its debuting DRIVE Thor platform is designed to “centralize autonomous and assisted driving,” following a trend in deemphasizing fully driverless vehicles.  The product is supposed to “replace numerous chips and cables in the car and bring down the overall system cost.”  Expect other companies to position their merchandise similarly.

Back we go to the New York Times, where Cade Metz again, of all people, wasn’t happy to be “Stuck on the Streets of San Francisco in a Driverless Car” (September 28th).  His go in a GM Cruise vehicle seemed to provide a combination of driving “gingerly” and “cautiously” and “skidding to a stop in the middle of a crosswalk,” precipitating a pedestrian giving the empty driver’s seat a familiar one-finger gesture.  It didn’t end well, as “the car detected a possible accident and pulled over,” after which, though it was “a false alarm,” “the car wouldn’t budge” and his “ride was over.”  Sad that these things are still happening.

The worst, though, was a, October 9th piece in Futurism, claiming that the “Godfather of Self-Driving Cars Says the Tech Is Going Nowhere.”  That view was from Anthony Levandowski, who founded a pertinent Google division, now Waymo, as one of its “key engineers.”  Among his non-cheerleading statements were “you’d be hard-pressed to find another industry that’s invested so many dollars in R&D and that has delivered so little,” “the industry still amounts to little more than a bunch of glorified tech demos (paraphrased),” “it’s an illusion,” and “why are we driving around, testing technology and creating additional risks, without actually delivering anything of value?”  The time and money, described elsewhere in the piece as “nearly twenty years and some $100 billion,” haven’t been lacking, but have solved neither the general problem of not knowing when to ignore “slight changes in the environment” nor the specific one of not effectively executing “the elementary move of cutting left across traffic when there’s no light to make it easy.”

Is Levandowski, who had serious problems within the industry later in his career, correct?  If not, we need to see autonomous vehicles indefinitely usable beyond the smallest and most protected of niches.  If not, the action will continue to be in technology that helps drivers, who, along with the jobs of the professional ones, aren’t going away.

No comments:

Post a Comment