August 2, 2017
Autonomous Vehicles: Truly Imminent?
by Bill Wilson
Are autonomous vehicles inevitable? Probably. Are they imminent? Probably not, at least on a widescale basis.
Just from casual observation, if “disruption” is the hot topic in industry media, self-driving or autonomous vehicles might not be far behind. I’ve read predictions that a proliferation of autonomous vehicles will be among us (or vice versa) in five to 10 years. Is that likely? I have my doubts, though I’m increasingly persuaded that the day is coming, even if it’s not clear how or to what extent the transition to autonomous vehicles will transpire and what impact it might have on our industry.
Fortune magazine published an article titled, “What’s Taking So Long for Driverless Cars to Go Mainstream?” The article notes that “significant technical challenges remain unsolved” but opines that “societal friction” will be the greatest impediment to adoption, delaying full autonomy “for at least a decade.” I think that length of delay is highly optimistic. There is a LinkedIn discussion about this here if you want to join the dialog.
The article discusses three societal impediments that could slow adoption:
- Ethics. If the vehicle recognizes that a wreck cannot be avoided and the choices are limited to running off a cliff and killing the vehicle occupants or ramming through a crowd of children, how will that decision be made, and who is going to be legally responsible for the consequences, if anyone?
- Job Losses. The article says there are four million jobs involving the transportation of people and property, not counting all of the support positions, along with other businesses that are transportation-related and dependent on human operators.
- Hacking. What are the possibilities of “weaponizing” wired vehicles? I’ve been writing about this for several years. Given the media attention to hacks of wired vehicles already taking place to illustrate vulnerabilities, what would happen if terrorists hacked into a gasoline tanker truck and sent it at 70 mph into an elementary school? With the “Internet of Things” (perhaps #3 on the list of industry hot topics), we well know the many vulnerabilities of everything from smart phones to internet-connected HVAC controls.
These are just three among many, many issues that must be addressed before ANY significant autonomous vehicle traffic is permitted. The article also cites reports that an Uber driverless vehicle ran a number of red lights last year. Aside from the loss potential, when that happens, who gets the law-enforcement citation? The vehicle? If someone is injured or property is damaged, who gets sued? Today, the vast majority of claims are due to driver error. When the inevitable autonomous vehicle claims occur, who gets sued…fleet owners, vehicle manufacturers, mechanics, software programmers or others we can’t even conceive of now? How will this change our legal system?
See also: Driverless Vehicles: Brace for Impact
Another article that challenges predictions that we are no more than a decade away from significant autonomous vehicle use is “Self-Driving Cars: Flawless Ride? Carmageddon?” Questions it raises include: Does the entire roadway system need to be re-engineered to accommodate both autonomous autos and human-operated autos? How will millions of miles or roads and an inventory of conditions, signs, laws and special situations be kept up to date and communicated to vehicle systems? How will decisions be made when accidents are inevitable? What about software bugs?
Questions I’ve raised in past writings include: How will an autonomous vehicle “know” that a traffic light is out? How will it detect a red light if the direct glare of the setting sun washes out the color? Will traffic equipment need to be re-engineered, and at what cost? How will it be tested and maintained? How will driverless vehicles respond to unanticipated road construction or weather conditions? How will they respond to circumstances not contemplated by their programming? With regard to software bugs, “rebooting” your car, unlike your PC or tablet, while moving at 70 mph may not be an option.
Fellow insurance nerd Mike Edwards told me about driving through his neighborhood one day and noticing, by glancing under the chassis, several sets of little feet behind a mini-van parked in a driveway near the street. Although the posted speed limit was 30 mph, he slowed almost to a stop just in case a child ran into the street…which happened. Fortunately, he was able to just stop in time. Could an autonomous vehicle be programmed with that kind of awareness? I’m certainly not an expert, but I have my doubts. Multiply this kind of incident several million times a year.
Finally, one of the more interesting commentators on autonomous vehicles is Thomas Frey, futurist at the DaVinci Institute and, according to his bio, Google’s top-rated futurist speaker. Here is a LinkedIn Pulse article he wrote:
His predictions include four million autonomous cars replacing 50% of all commuter traffic. In a city of two million people, only 30,000 autonomous cars might be sufficient to handle transportation needs. It’s still not clear to me how 150,000 people leaving the vicinity of an Ohio State vs. Michigan football game will be able to access a limited number of largely single- or dual-passenger autonomous vehicles in Columbus or Ann Arbor. I’m not even convinced that autonomous vehicles can effectively handle rush hour transportation without there being tens of thousands of vehicles idle during non-peak hours. Will people willingly share vehicles with strangers at close quarters, and how might that contribute to crimes against persons? Will people be OK with taking twice as long to get somewhere with slower traffic and possibly multiple drop-offs like airport shuttles today?
Under most current laws, as much as 40% of state and local sales taxes could be lost if individual auto sales become a thing of the past…what are the funding alternatives? The same issue surrounds state and federal fuel tax revenue.
Remarkably, Frey predicts that, within 30 years, fully 25% of current jobs will be lost due to autonomous vehicles (e.g., he says 80% of police manpower goes to traffic control and will largely be unnecessary). New York City would allegedly lose more than $2 billion in income from traffic fines alone. Airport revenues would plummet, given the claim that 41% of it comes from parking and ground transportation fees.
On the upside, according to the author, we would save $500 BILLION annually in healthcare costs (though two other sources site the negative impact of fewer organ donations due to fewer auto accidents).
The list, good and bad, goes on and on…take a look at his predictions at the link above, including one for the insurance industry.
Are we moving not to a “sharing” economy but to a “hiring” or “renting” economy where there is little in the way of ownership of high-ticket items like homes or autos?
Perhaps, though I’m not sure any of us can predict what is going to happen with any degree of certainty or say how soon it will happen.
The closest large-scale implementation of semi-autonomous vehicles so far is probably our aviation system. Air transportation is highly automated, from take-off to landing, but the pilot(s) are able to take control if necessary. Even so, traffic control of aircraft is only marginally automated. Humans often prevent tragedy through the oversight, coordination and intervention of air traffic control centers all over the country with individual aircraft. This system has demonstrated itself to be quite effective, given the very low incidence of casualties associated with commercial flight.
See also: The Evolution in Self-Driving Vehicles
But this centralized, coordinated control aspect may be largely responsible for the success of the system. Unlike autos, no one is currently advocating that individual aircraft become autonomous. That’s interesting given that the number of aircraft in the air at any given time is tiny compared with the number of vehicles on the road, and probably no one would argue that there are a lot fewer structures and people to hit in the sky vs. a congested city. So, what makes the pundits believe that tens of thousands of individual autos can coordinate themselves effectively without some sort of centralized control or the ability to communicate between each other?
Currently, when a driver makes a mistake or wrong decision, the adverse results are limited to a single accident, usually with one or a few vehicles involved. If a programmer makes a mistake or wrong decision, the adverse results could be astronomically higher.
Are autonomous vehicles inevitable? Probably. Are they imminent? Probably not, at least on a widescale basis. What do you think? As always, feel free to add your thoughts in the Comments section below.