It’s been seven months since my last post on self-driving automobiles. Normally, that’s too short a time to tackle the subject again even for many technical propositions, but this one has been zooming along like a Ferrari. Since November, a mass of work, most being completely proprietary to the software companies, automakers, and electronics firms doing it, has been finished. The otherwise beleaguered gypsy cab companies Lyft and then Uber have been working with Ford and General Motors toward offering driverless taxis. Microsoft and Alphabet, Google’s parent company, have made their like ambitions public. An elaborate artificial city section in Ann Arbor, Michigan, complete with storefront facades, 13 different traffic light varieties, expressway ramps, and parking meters, is being used by at least five world-class car manufacturers to test their autonomous products, and others are under construction in Ypsilanti, California, and Virginia. There has been an outpouring of articles, ranging from Neal E. Boudette’s New York Times “5 Things That Give Self-Driving Cars Headaches” to one of course much better than the grammar of its headline, “As a senior citizen, a self-driving car will be my godsend,” in the Washington Post, and on to attempts to find the best related investment targets. There is also a subtopic taking form about driverless large trucks, with many companies seeing them as a way of resolving their long-standing driver shortages without sufficiently raising pay.
Although it is impossible to even decently estimate the amount of money and person-hours currently going toward this effort, it is crystal-clear that by 2020 or so such technology will be usable for ordinary consumers. That, though, will prove to be its least limiting factor.
The largest sticking point is the lack of a suitable environment. At the least, places allowing driverless vehicles will need large-scale Wi-Fi, for the constant communication with online facilities these cars and trucks will require. Citywide or better still statewide or nationwide wireless connectivity, built robustly and redundantly enough to make outages almost nonexistent, is the best way of dealing with potholes (located precisely through GPS), poor visibility on snowy roads (through detailed maps), and detours (with up-to-the-minute rerouting instructions). Unfortunately, that is exactly the kind of infrastructure project that Congress has refused to fund. Ultimately, only countries willing to build such a network will be able to fully implement driverless technology.
There is a set of additional problems which might be described as barriers to mentality. The first is one of the five headaches in Boudette’s story, described as “ethics on the road.” One example he poses is a ball, followed by children, bouncing into the car’s path, giving it no choices beyond hitting the ball’s pursuers or crashing into something on the side of the road and endangering its occupants. In this and many other situations, some people will lose and others will be spared, and there will be no human driver to explain his or her actions, only computer code with unknown originators. Another is bizarre accidents rare but not unheard of, which will kill people before they are programmatically forestalled. A third problem is one more from Boudette, “unpredictable humans,” which may never be eliminated. (Could a modern-day superbly maintained bullet train driven on its immaculate track by a completely sober, prudent, and experienced engineer avoid tragedy when a person, suicidal or otherwise, somehow appears, say, 130 feet directly in front of it, or half a second away at its 180 miles per hour?) The fourth issue concerns autonomous 18-wheel trucks, of which many people will, even if mishap rates are microscopic, be terrified. Although it is clear that driverless vehicles will prove vastly less dangerous than human-driven ones, resistance to the accidents they will have, many of which would not take place if they were manually driven, will still be at least a temporary problem.
What are the chances that we can overcome the concerns above? How might self-driving technology be implemented in stages? What else will happen with and around it? These will be the subjects of next week’s post.