The past eight installments of this series covered a lot of ground -- some technical, but mostly socio-economic and conceptual. But in the debate over the pros and cons of human versus robotic drivers, is it not possible to have the wisdom to take the best of both worlds?
In early articles I acknowledged some of the advantages of Highly-Automated Vehicles (HAVs), including:
- No fatigue. No need for layovers. Costs for OSA testing will vanish. And we can remain largely ignorant about bio-sensitive scheduling and driver assignment. Problems like shift inversion will vanish overnight by definition.
- No reaction time. Vehicles will stop a lot shorter than they now do, when situations warrant them to. (At the same time, I suspect the motoring public will not understand this, and rear-enders will pile up like foreclosures.)
- No training. Training will be replaced by programming. Yet while programs will likely become increasingly obsolete, and systems may occasionally crash, nothing will be forgotten. Back-ups are forever.
- No drivers to pay (along with their fringes, employment taxes and pensions). Concepts like recovery time will vanish -- even though deadhead time and mileage will still matter.
- No human knowledge or memory. To the degree they can make judgments -- with questionable limits to this in my opinion -- the robots will have an encyclopedic command of information. Mapping software like google.com/earth is only a crude starting point. Notwithstanding some need for judgment, negligent policy decisions and poor roadway design, principles of defensive driving, square turns and countless other operating details will be retained and applied by the vehicles' brains.
- More roadway capacity. With no reaction time combined with the ability of a vehicle to see further ahead, following distances will decrease for any given speed, and roadway capacity will increase as a consequence.
- No need to kick tires, conduct pre- or post-trip inspections or read gauges. Someone or something in charge will get all the data instantaneously. Of course, the decision to respond to what these data reveal is another matter that will lie beyond the robots' control.
At the same time, if current domestic experiments are any indication, it will be a long time before the HAVs supplant the judgment of our finest drivers. Sharp humans can likely recognize nuances that robots cannot. At least not in the foreseeable future. Without such clues, how will a camera or sensor know how long to linger on an object, or how quickly to return to it? There is a great jazz tune from an obscure movie of the same name: "The Night Has a Thousand Eyes." Will an HAV need 500? Even if they see everything, how accurately will they respond to it all? I do not foresee robots sensing body language, although I suspect that natural facial expressions can be catalogued. But will a robot be able to tell if some pedestrian "makes a face?" Would Tomorrow's terrorists merely need to become good comedians or actors?
In Romeo and Juliet, Montague challenged Capulet, "Do you bite your thumb at me, Sir?" Capulet responded, "No, Sir, but I do bite my thumb sir." Life has always had its gray areas. So too will the robots. I suspect it will not take terrorists or hackers long to master them.
Muddy Waters and Delusions of On-Time Performance
No moving vehicle exists apart from its duty cycle and operating environment. Some of our waters are still uncharted. As an economic matter, our infrastructure is dysfunctional beyond any hope of repair, and we face a future of detours. What solution to this problem can a driverless vehicle contribute? Investment dollars are already pouring into startups for flying cars and roadable aircraft. Particularly given an HAV's higher costs, how much more can it offer us given our infrastructure constraints?
Then there are the tight schedules -- particularly in the transit, paratransit and, worse, non-emergency medical transportation sectors. How can an HAV with a long wheelbase make a proper turn when the schedule is already tight despite scores of other safety compromises that still cannot help the route run on time? Within the constraints of our operating environment, and the budgets and policy decisions that limit it, HAVs will encounter challenges to what they can accomplish in terms of safety, irrespective of their technical capabilities to do otherwise.
Then there is the incompatibility of our street network with vehicles with long wheelbases. HAVs are not going to re-engineer and re-construct our vast roadway network with wider streets, rounded corners and properly-located limit lines. We cannot afford to maintain those we already have. Yet even if we could afford to make the changes needed, do we actually expect our decision-makers and technical experts to take policy advice and engineering judgment from robots?
The Nature of Law Suits and Litigation
One initial reaction (expressed in previous installments of this series) is that plaintiff's attorneys will fall decades behind in their ability to hold an HAV or its suppliers accountable. For starters, it should be easy to find countless ways in which HAVs' accident rates are favorable to those of vehicles driven by humans. Plus, it will take years to create a sample size large enough to even make such comparisons -- and the slower HAVs dominate the transportation landscape, the longer it will take for a valid database for comparable purposes to materialize. So while HAVs are indeed inevitable, I do not expect them to show up very soon, much less beneath this year's Christmas tree.
At the same time, the type of failures HAVs experience are likely to be quite different than those committed by human drivers, and there should be plenty of expert witnesses who can identify the superiorities of HAVs convincingly. But these individuals, too, may take a while to emerge and develop. Plus, troubling trends in the litigation field will give the HAVs and their suppliers increasing latitude, and even more time to adjust.
In Control and in Charge
The differences between these two constructs may seem subtle, but they are not. Unless we allow the robots to make policy decisions, they will necessarily be constrained significantly by many of the errors and omissions that live Earthlings will continue to commit. Robots commandeering moving vehicles can surely exercise superior control over many, many things. But are we really going to put them in charge of determining what things will lie under their control?
It is also important to recognize that while both ships and aircraft are largely piloted by robotics (and have been for decades), they both still have crews. There are moments when human judgment is needed. Calling the command center in Secaucus, N.J. to second-guess an NBA referee's call-on-the-floor is not an operable model in the world of large, heavy, fast-moving objects with a lot at risk, where near-instant judgments are often needed, and post-incident second-guessing is of marginal value (outside of lawsuits).
And then there are U.S. roads, bridges, tunnels and other infrastructure. Japan had high-speed passenger rail since 1992, and by 2015, it was traveling 177 mph. The Acela's top speed, between New York City and Washington, D.C., is 135 mph -- and it cost tens of billions to upgrade our roadbeds to even facilitate this speed. We cannot remotely afford to upgrade our roadbeds to accommodate European and Asian passenger rail speeds. The same is true of Autobahn-like freeways. Driverless buses and coaches will be trapped in this same substandard infrastructure, notwithstanding tens of thousands of detours likely to be in place by the time an autonomous bus or coach reaches our shores. Yet I suspect they can adjust speeds and fuel efficiency on all those gravelly country roads that used to be paved.
Otherwise, I expect some of the simpler technologies of driverless European buses and coaches to trickle down to our operating environment to simply expand upon the limits of safety that our live drivers and their management and policy-makers can provide. And I expect these efforts to quickly surpass some of the bungling efforts cited in installment #7 of this series (see "Part 7, Autonomous and Inevitable, Part 7: Cameras and Sensors," in National Bus Trader, May, 2017).
Also, it is hard to know whether or not the elimination of drivers will offset the higher capital costs of advanced-technology robotic buses and coaches. (Will the FTA still pay for 80% of the new bells and whistles on transit and paratransit vehicles?) Perhaps the difference will be revealed in damage awards and law suit settlements. Otherwise, the comparison of higher capital costs compared to lower operating costs is hard to predict. But what is not hard to predict is the unlikely increase in fares to cover any cost increases that may emerge.
Finally, let us place autonomous ground vehicles in perspective. This past year I have been overwhelmed by the concerns of such vehicles being hacked -- perhaps not totally unjustified, since, until recent years, roughly the target of half of all terrorist attacks have been buses. Yet we do not seem to be concerned about this with our ships and aircraft -- even while some huge tankers have been kidnapped by pirates. But our ships and aircraft are immune to hacking because they have live Earthlings at or near the controls, who can take over on a moment's notice, and have enough analog skills to master the safety-related tasks they would face if their digitalia crashes.
So we would be fools to simply shrug off the value of drivers, even if their tasks are minimized. Instead, we should focus on how total vehicle operations and control could improve by combining live, trained drivers (perhaps trained differently) with all the artificial intelligence we can shoehorn into their vehicles, and all the things (see short list above) they can accomplish. Introducing robots into the dialog may challenge the cliché, but frankly, the best individual can rarely beat a team. The team-up of a good humanoid and wisely and prudently-selected and -tasked robots would seem to be an impossible team to beat in terms of both safety and security.
Balance, Pace and Compromise
It would be foolish and naive to consider HAVs an all-or-nothing alternative. While the corporate American dream of eliminating drivers may eventually come to pass, we should not give zero thought to integrating the best that both worlds have to offer. Some mid-price passenger cars are already doing this. This current integration is certainly not the ultimate endpoint reachable. But it seems like a better direction to pursue during the development and application of new technologies than a simple competition.
In this debate, it is also important not to panic. The pace of change is clearly happening at break-neck speed. But not everywhere. And like it or not, we are not Europe or Asia. China is about to build a 260-mile passenger rail system that travels through eight countries for an estimated $6B. Our intelligentsia recently spent $4B to unnecessarily upgrade a single rail station (the World Trade Center PATH Station) with a maze of shops so expensive that most passengers cannot afford to even window shop alongside them. They also have less time to do so since the renovation quadrupled street-to-platform pedestrian travel time. Today's America is a long way from the Jetsons.
I welcome new technologies for the promises they offer. But I also do not accept compromised safety, unemployment and other dysfunctional consequences as inevitable during the transition. And I think the speed of the vision materializing about HAVs has been grossly underestimated. This may not be such a bad thing: If we do not squander it, this delay may give us time to sort out the competitive advantages of humans versus robots gradually, with both safety and socio-economic impacts given a chance to compete with corporate commercial agendas. More importantly, it might give us a chance to combine the best of both Worlds.
Under such conditions, I welcome the robots. I just welcome them more slowly, more rationally, more responsibly, more thoughtfully and more wisely.