Wednesday, November 22, 2017

Driverless Cars – More Progress and Positioning – VI

To start out, “Daimler shows off what its automated Mercedes-Benz Arocs trucks can do” (Darrell Etherington, TechCrunch, October 17th).  That isn’t shabby.  Like good soldiers, these German-made “industrial vehicles” gather in formations, can each take the lead, and follow “a strictly laid out and mapped path to coordinate their efforts.” 

They’re also taking Manhattan, or five square miles of it anyway, now that we can expect “GM to Test Fleet of Self-Driving Cars in New York” (Mike Colias and Tim Higgins, The Wall Street Journal, also October 17th).  We don’t have an expected date, but we do get the Boston Consulting Group projection that 25% of American-traversed miles could be provided by “shared, self-driving vehicles” in 2030, the insight that “hard-earned city driving is more useful for the car to learn how to handle unusual situations that human drivers take for granted, such as how to handle broken traffic lights at an intersection,” and Cruise Automation CEO Kyle Vogt’s view that this city will “present some unique challenges.”  The latter is exaggerated, but autonomous vehicles will indeed need to learn how to handle aggressive human drivers, and New York is almost as good an American place as any to do that. 

One thing hawked by developers of driverless vehicles has been their projected effect on road congestion.  “Self-driving cars could ease traffic, but increase sprawl” (Fox Business, still October 17th) reported on another Boston Consulting Group finding, that the technology could “likely add vehicles to roads while simultaneously reducing traffic time.”  Indeed, both should happen, though it may be nearer the end of the driverless revolution than the beginning.

We discovered another joint venture in “LG, Qualcomm join hands for autonomous driving” (Fox Business again, October 19th).  This one brings in another Korean behemoth company, this time working with an American chip manufacturer to develop “fifth-generation wireless technology known as 5G,” which “is seen as crucial for autonomous vehicles.”  The work will be done in Seoul, a good idea since South Korea, with its compact size and high connectivity standards, could quickly find itself on this forefront. 

One story, published only days later in the same place as a doubting editorial, may have offset, if you want to call it that, its damage.  A firsthand report from David Leonhardt, an established columnist there, The New York Times’s October 22nd “Driverless Cars Made Me Nervous.  Then I Tried One,” related his experience going through suburban Washington in a “semi-driverless” Volvo S90.  Leonhardt related how, even though “we don’t want computers to be in charge,” this sedan surprisingly, to him anyway, behaved itself.  Judging by his inserting talking points such as the 30,000+ toll of American driver errors, the strongly positive effect of “automation” on reducing airline crashes, and the already-established tendency of driverless accidents to be “sensationalized… while we ignore tens of thousands of deaths from human crashes,” David Leonhardt has been won over, and we can expect many more to follow.

Also on October 22nd, The Motley Fool stepped away from its usual investment emphasis to opine that “Google’s Strategy to Educate the Public About Self-Driving Cars is Brilliant.”  Author Danny Vena reminded us about Waymo’s huge-for-the-industry eight years of autonomous vehicle development, and told us some bits and pieces about its “public education campaign,” which, with the amount of concern growing about the technology, is a necessary part.

“I Just Drove The Car Of The Future – And It Wasn’t A Tesla.”  So reported Thomas Koulopoulos in on the same date.  He identified himself as having a “lifetime love of driving,” a sentiment not usually consistent with enthusiasm about autonomous cars, but he seems sold on the technology.  Among his intriguing views were that, instead of 90% of it, “the entire fleet of global transport will switch over to driverless with(in) 25 years,” that “gateway innovations” such as cruise control and lane departure warnings are “slowly weaning us off driving,” that in order to “acclimate drivers to being passengers” the companies will need to make “many incremental changes,” and that people will someday react to our early-21st-century driving experiences comparably to those today seeing “an old stagecoach or horse drawn wagon” and wondering “how in the world did people ride in those things for more than a few miles without having their teeth shaken loose from the jarring of wooden wheels on unpaved roads littered with rocks and ditches?”, when those traveling thusly knew that method was, in their time, “advanced.” 

Another piece part, one of many that may turn out to be invaluable for driverless vehicles, was acclaimed in “Comma AI’s dash cams are a stepping-stone to autonomous driving” (Roberto Baldwin, Engadget, October 24th).  Though these devices may seem unrevolutionary, as they are only OnePlus 3 Android phones with other things added, they still provide video records of trips, including any accidents.  Also worthy technology is the subject of the same author’s and source’s October 26th “Baidu updates its open-source autonomous driving platform,” which this Chinese driverless company expects to “be powerful enough for a Level 3 car on the road by 2019, and primed for (a) Level 4 vehicle in 2021.”  All from a firm not only “not even building cars,” and located in a country where mapping problems make it unlikely to achieve its autonomous-vehicle potential.  Yes, it is a worldwide industry, and one where United States companies will need to continue to work hard to maintain their lead, which they, and others, seem thoroughly willing to do.  That will prove to be an excellent thing.  

Friday, November 17, 2017

Driverless Cars – More Progress and Positioning – V

In this series I’ve been writing about how Ford’s autonomous vehicle progress has been better than General Motors’s.  Now that “GM Enhances Self-Driving Car Effort With Deal for Strobe” (Mike Colias, The Wall Street Journal, October 9th), is that still true?  Strobe made lidar, a critical driverless technology that allows such vehicles to see what is around them, but was small enough to have only 11 employees.  Will this corporate behemoth effectively manage this entrepreneurial startup without crushing it?  Or does it just want the know-how Strobe has already accumulated?  In any event, even if Ford and GM end up having a seesaw battle, this acquisition isn’t enough to turn the latter company around. 

In partnerships of another sort, “Waymo teams with MADD, the NSC and more on self-driving education” (Darrell Etherington, TechCrunch, October 9th), the forerunning driverless nameplate has not only joined the National Safety Council, but has found common cause with three social organizations, the Foundation for Blind Children and the Foundation for Senior Living as well as Mothers Against Drunk Driving.  All three are natural supporters of autonomous vehicles, and can help Waymo in getting political support, something they may end up needing a great deal.  Another winner for them, as was its release three days later of its “Safety Report” (same author and source); the 42-page piece for public consumption explains how autonomous vehicles work, their current cybersecurity defenses, and “what happens after a crash should there be one.” 

Speaking of winners and good progress, the online ink was barely dry on The Motley Fool’s recommendation of this company’s stock, when “NVIDIA introduces a computer for level 5 autonomous cars” (Roberto Baldwin, Engadget, October 10th).  The device, “the size of a license plate,” “delivers 320 trillion operations per second, 10 times more than its predecessor,” but even NVIDIA admitted that it won’t be put into service soon, and then maybe only for “robotaxis,” on which “over 25 of its partners are already working.”  When it powers an NVIDIA “pilot fleet” next year, we will know more.   

Yes, it is a certainty that “Now is the time to plan for the autonomous vehicle future” (Tom Alberg and Craig Mundle, TechCrunch, October 11th).  We learn herein that Audi, from whom we have heard little, now plans on “selling, in 2018, a production car with Level 3 authority (meaning it requires no human attention to the road at speeds under 37 miles per hour).”  That just might be the first one available for general purchase.  Alberg and Mundle also stated that a proposal has been made to the Seattle government to convert Interstate 5, which runs to the Canadian border on the way to Vancouver, to driverless-car-only use, with an intermediate stage of changing high-density vehicle lanes into autonomous-vehicle ones.  Although many of the authors’ talking points are weak – it’s not clear that driverless cars will lead to “less congestion, reduced emissions, ...fewer new roads, reclaimed parking space,” or “lower transportation costs for all” – we must agree with the article’s ending, that “it is time to get going.”  

Also, “California will allow self-driving cars with no human driver on public roads next year” (Sean Szymkowski, MotorAuthority, October 12th).  The door is now open for more space in that state than Silicon Valley.  Three years out, “Baidu plans to mass-produce Level 4 self-driving cars with BAIC by 2021” (Darrell Etherington in TechCrunch again, October 13th).  Baidu, a huge Chinese company expanding from their Internet-providing core business to driverless technology, will find its most formidable adversary to be its country’s government, which has hampered other companies by impeding map updates, so we’re not sure if they will stay on schedule at all.

With all the positive news above, we were probably due to see a cold-water-splashing major-newspaper editorial, and we got one on October 14thThe New York Times Editorial Board spent its main Sunday space on “Would You Buy a Self-Driving Future From These Guys?”.  It has present/future confusion, with “87 percent favored requiring that a person always be behind the wheel, ready to take control if something goes wrong” (fine for 2017, but that number will be way down by 2020 and we’ll laugh about that viewpoint in 2050), yet more mention of the May 2016 fatal Florida accident (since then, over 45,000 Americans have died in driver-caused crashes, which the Florida one was determined to be also), attempted talking points of “mass unemployment for taxi drivers” and “greatly reduce(d) car sales” (veterinarians and blacksmiths lost jobs also, in the course of the last major round of ground-transportation progress and almost immeasurable prosperity enhancement).  True, auto safety standards should remain, but those connected with driverless vehicles are not, as the headline implies, generally malevolent, and assuming they are would in this case be wicked and destructive in itself.

By the time these cars are in general use they will be safer than they are now.  But for the time being, there is, as with everything else in this area, a lot of work needed.  One good thing about “Low-speed accidents frequent among driverless cars” (Ryan Beene, Bloomberg News, October 15th), is the hyphen-connected words starting its headline.  The root cause of most of these crashes is that, per Gartner analyst Mike Ramsey in the article, autonomous vehicles “don’t drive like people.  They drive like robots.”  We, though, will be prepared for that, through everything from new driver’s-education teaching through almost certain changes in our own driving behavior, and will adjust.  Also, as above, the cars do indeed need more development time, and are getting it.

Eventually, this series will end.  But in the meantime, expect yet more next week.

Friday, November 10, 2017

Driverless Cars – More Progress and Positioning – IV

One area of autonomous vehicle research much noted over the past two months is their proving facilities.  The country’s second-largest automaker returned to the self-driving headlines therein with “Ford invests in Michigan’s autonomous car testing grounds” (Timothy J. Seppala, Engadget, September 15), showing that it will likely not be behind General Motors in this next huge phase, and that the state of Michigan, for similar reasons, will also make a comeback.   

Another thread in this area has been consumer reactions.  In Salon’s September 16th “Lots of love for driverless cars, just not from drivers,” Paul Feldman recapped survey results showing “that most motorists don’t want to drive, ride in or be on the road anywhere near” autonomous vehicles.  With their current technology that is appropriate, and indeed it is not even possible yet for ordinary people to use them outside of a few free-ride programs.  With reporters seeing firsthand and later publicizing how well such cars fared with impediments last month, that confidence level has since improved and will continue to rise.

One more driverless domain has been the completed and expected 
implementation of their ever-improving features in meatmobiles.  As Lee Vinsel put it in the September 23rd “The best parts about autonomous vehicles are already here,” also in Salon, “elements of self-driving car systems, such as adaptive cruise control, lane-departure warnings and head-on collision-avoidance systems, could reduce road deaths by up to one-third.”  That would be over 11,000 in the United States alone, without any benefit from the vehicles for which these technologies are eventually intended.  This article also included the historical insight that the federal National Research Council “became interested in a self-driving car” in 1953, and that Vladimir Zworykin, often credited as the inventor of television, was doing much of the research. 

“Could You Form an Emotional Bond With a Self-Driving Car?”  My response to Doug Newcomb’s September 23rd PC Magazine article is, well, maybe, but nowhere near as much as when such connections, probably in the 1950s or 1960s, reached their peaks.  Personal intensity about vehicles has been falling ever since then, is either a cause or an effect of the drop of young people getting driver’s licenses over the past decade or two, and, as interesting as these ideas are, won’t be reversed by synchronizing music with driving speed or installing “a “hunger undulator” that uses vibrating motors to mimic stomach contractions so passengers feel hungry as the car starts to get low on fuel.”  True, people can get attached to machines, from Star Wars droids to copiers, but it’s hard to see how that will reach its former strength.

Back to the business side, with “Ford and Lyft Sign Driverless-Car Agreement” (Greg Bensinger, The Wall Street Journal, September 27th).  Here’s where Lyft, now more respected, could become larger than Uber, as it has also been working with General Motors.  The latter company may or may not be getting anywhere in the area itself, per “GM making ‘rapid progress’ toward self-driving car development: executive” (Reuters, October 3), a piece which, based on GM press releases, offered no real specifics. 

Some retirement-oriented places, in particular Sun City West in Arizona, have allowed golf courts on streets for decades now.  Daisuke Wakabayashi, in The New York Times’s October 4th “Where Driverless Cars Brake for Golf Carts” describes how the Villages Golf and Country Club in San Jose has made its vehicle mix even more diverse, with 15 miles of roads, a 25 MPH speed limit, and a private-property status making it “an ideal proving ground.”  Community management sought a lower-cost way of providing shuttle services, found that residents actively wanted driverless cars, and connected with Udacity, which is now there testing automated vehicles of some sort.  This Villages implementation may go down as the first time driverless technology was sought out, and succeeded, to save money.

On the regulatory side, the Associated Press reported in Fox Business on October 4th that “Senators weigh bill to remove obstacles to self-driving cars.”  We’ve seen the controversy behind regulatory issues before, and it’s the same here, with one side citing a potential end to 94% of car accidents and the other, expressed by a onetime leader of the National Highway Traffic Safety Administration, saying that “the public will be the crash dummies in this experiment.”  Both sides need to be heard, but I am confident that, as technology and familiarity both improve, the anti-autonomous faction will become smaller and more of a fringe.

One odd characteristic of driverless vehicles of all kinds is that, opposite to those human-operated, they are less dangerous when more of them are on the road.  In Engadget’s October 4th “Networked self-driving cars are smarter and safer,” Daniel Cooper explained why.  “Two cars in convoy” perforce occupy different spaces with different fields of vision, so if they can combine what they see it will be more than from only one.  That’s progress, as is the subject of the same date’s “General Motors Just Made a Lot More Self-Driving Cars” (Reuters).  Though “a lot” is only 100, and though as before we get few hard facts about GM’s work, it’s still movement in the right direction.

Clearly, the headline of Srikanth Saripalli’s October 8th Salon piece “Are self-driving cars the future of mobility for disabled people?” deserves a three-letter answer.  And the same is true for drinkers.  The article correctly points out that there will be logistical issues to resolve, such as “identifying curb cuts that let wheelchairs and walkers pass easily as well as noting potential obstacles, like trash cans out for collection,” but, indeed, driverless vehicles “have the potential to change neighborhoods and individuals’ lives – including people who are disabled and often both literally and figuratively left behind.”  That should serve as an offset to those pessimistic about safety. 

We wrap up this week with a heady philosophical issue, in “The Breakthrough We Need for Self-Driving Cars to Become a Reality” (Lily Elefteriadou, Popular Mechanics, also October 8th).  We need not care about the main point of the article, which is publicizing the University of Florida’s good but elsewhere-duplicated research, but is it true that “no consumer wants to buy a self-driving car that’s programmed, even in the most remote of circumstances, to kill its (occupant)”?  Either fortunately or unfortunately, when algorithms are developed to make life-or-death decisions when only destructive actions are possible, it may be the law for automated vehicles to crash instead of hitting a crowd of pedestrians.  That would be an intriguing premise for a science-fiction story, with, as so many have, real-life application, involving illicit reprogramming counterhacked by federal forces.  Yet it is nonfiction that the cars themselves are on the way.

More to follow next week.

Friday, November 3, 2017

October’s Mixed Jobs Report: American Job Shortage Number (AJSN) Drops 500,000 to 16.2 Million

I was surprised by this morning’s Bureau of Labor Statistics employment figures.  Were you?

I wrote last month that September’s data was mangled by the storms, giving us strange results such as the loss of 33,000 net new jobs.  Accordingly, we would need to combine it with October’s to get a fairer view of how our working economy is really doing.  So, what happened?

In October, we gained 261,000 net new nonfarm payroll positions.  That would normally be outstanding, but when we combine it with the last set and divide it by two we get 114,000 per month.  With population increase consuming over 130,000, that’s a slowdown.  Also on the down side, the two measures of how common it is for Americans to be working, the employment-population ratio and the labor force participation rate, erased and more than erased respectively their September improvements, going off 0.2% and a steep 0.4% to reach 60.2% and 62.7%.  Average private nonfarm payroll hourly earnings gave up a little of its large September improvement, down 2 cents per hour to $26.53. 

Many of the other numbers were unexpectedly good.  The adjusted unemployment rate fell 0.1% to reach another post-Great-Recession low of 4.1%.  The unadjusted variety, lower since October is typically an above-average working month, achieved the same, shedding 0.2% to get to 3.9%.  These two numbers were also down, 0.2% and 0.4%, in September.  Those officially jobless and out for 27 weeks or longer cut another 100,000 to reach 1.6 million, and the count of people working part-time for economic reasons, or seeking full-time work while holding on to shorter-hours employment, plunged over 300,000 to 4.8 million.  In September the former number stayed the same, but the latter one, which usually changes half that amount or not at all, improved 200,000. 

The American Job Shortage Number or AJSN, which shows in one figure how many more positions could quickly be filled, dropped an unexpected half-million to reach another post-recession low, as follows:

The AJSN’s drop was only partially caused by lower unemployment, which accounted for 282,000 of the difference.  More significant was a cut of almost 400,000 in those wanting to work but not looking for it in the previous year.  Both were offset slightly by increases in those calling themselves discouraged and a surprising 1.2 million gain, precipitating a 60,000 AJSN difference, in the count of those claiming no interest in employment. 

Compared with a year before, important since the AJSN is not seasonally adjusted, it has dropped over 1.1 million, with almost that amount from lower official unemployment, and over 300,000 from the did-not-search-in-previous-year category partially neutralized by 1.2 million growth in those in the armed forces, in institutions, or off the grid, and by a 700,000 gain in the number of American-citizen expatriates now overseas.  The share of the AJSN coming from those officially jobless, now under 35%, reached another post-recession low, and means that almost two-thirds of those taking new positions would not be expected to be counted as unemployed.

Overall, how good was the data for the past two months?  Lukewarm.  Fewer people are unemployed, unemployed for six months or longer, or struggling with part-time positions instead of the full-time ones they want.  Wages are ahead of inflation.  Yet jobs growth and labor force participation are on the wane.  We’re only one number, specifically “did not search for work in previous year,” away from saying we’re going nowhere, and that work in America is continuing to depart, however slowly, from being the norm – and with that metric’s tendency to fluctuate without a long-term trend, that isn’t enough.  The turtle stayed put.