Sunday, September 11, 2016

The Future on Autopilot

Self-driving cars are not science-fiction. This generation will grapple with the societal implications of the automation of much of personal and commercial transportation.

Many desired technology innovations—personal jetpacks, thriving moon colonies, an iPhone that doesn't need to be charged daily (ha!)—may never be realized, but one is a lot closer than most people think: self-driving cars. While interning at Google this summer, I witnessed sister company X's autonomous vehicles on Mountain View roads and attended speaker events with people on the project (with which I had zero involvement), and have become convinced this is the future. The technology, while still improving, is indeed viable—a when, not if situation—and the changes it represents could improve personal safety, relieve congestion, lower transportation costs, and reduce environmental impact.

The most salient benefit of self-driving cars is that they will eventually be able to perform inarguably more safely than human drivers, whose errors account for hundreds of thousands of deaths around the world each year. Able to "communicate" with one another, self-driving cars will also maintain speed and handle relative positioning more efficiently than human drivers, improving traffic flow and changing development patterns. But there's much more to it. Currently, the cars we own remain unused most of the time, taking up space when they are not taking us from Point A to Point B. Optimally, a (clean emissions) vehicle would be in operation as much as possible, serving the needs of many commuters, so fewer people would require vehicles exclusively their own. Among the smaller pool of total cars needed to serve a population, those in use would be far less likely to idly occupy a parking spot. Much of the urban and suburban space currently dedicated to parking lots and garages could be repurposed for more productive ends, including green space.

If technological feasibility isn't a major barrier, regulatory and liability concerns remain enormous issues to work out, especially because people will be far less forgiving of machine error on the road than human error. Still, I was surprised to see the announcement that this very month Uber is deploying the first of 100 SUVs with autonomous driving capability in Pittsburgh, picking up and dropping off passengers through its app like it does with any traditional ride.1 For this pilot program, a human is still present on standby ready to take over whenever the computer needs assistance. This sort of system is what is called a "Level 3" capability, where a human driver must take control whenever alerted to do so by the computer.2 Impressive, but by Uber's own admission that trails behind what rival Google/X can do. The latter is working toward Level 4: no human involvement, where X's vehicle is a mobile pod sans steering wheel and pedals (a la Minority Report, pictured). Society is not ready for that level of comfort with the technology, nor must it be, but in a world of ever-decreasing cycle times for adoption of entirely new habits widespread Level 4 automation may be a long-term eventuality.

The most likely outcome of making predictions about the future is to look back later and laugh about how wildly inaccurate they were, but I'll take a stab anyway: Level 3 automation for use on highways—typically long and mostly straight stretches without the complexities of city streets—will be approved and adopted soon. The first major application of this technology will be commercial, though, particularly long-haul trucking transformed by systems (such as those being developed by Daimler, Peloton, and recent Uber acquisition Otto) that dramatically reduce the burden of human drivers. Then, public buses and the traditional taxi/rideshare fleet in major cities will be replaced by Level 3 and some Level 4 vehicles, which will also greatly expand home delivery services. Finally, at a much slower pace, autonomous cars will gain share in the personal transportation marketplace. For decades to come people will own traditional cars but at some point their continued existence in developed economies will be only for extremely complex routes (and to serve hobbyists). In the absence of human drivers to blame in an accident, liability will shift to the makers of autonomous vehicles—and enterprising businesses in this space will have to decide whether there is an advantage to producing entire cars or simply outsourcing the underlying autonomous technology to established car manufacturers.

Obviously, autonomous vehicles will bring major disruptions to multiple industries and affect everyday lives. Yet, for the earlier cited reasons of safety, environmental impact, and better land use, the disruptions if anticipated and managed properly will still yield a net positive outcome for society. Like with other areas which modern industrialization has transformed, certain jobs will no longer exist, other new jobs will be created, but handling that transition so it isn't socially destructive will require a mindful approach.3 More so than with other technology-affected areas, transportation is personal, and reactions to change here will be highly emotional. (The bitter fight over the past several years between cabbies and ride-sharing advocates will seem downright pleasant in comparison!) As The Economist drily noted earlier this month, "Offering cheap rides is not going to be enough to count on the public’s good graces."

1. That's well beyond the "Level 2" capability of assisted-driving features such as those seen in a Tesla, where a computer handles multiple "primary control functions" such as adaptive cruise control and lane centering, but the human driver should still man the wheel at all times. The tragic fatality in May of a Tesla driver using the car's (inaccurately named) "Autopilot" system occurred when a white tractor-trailer was not detected against a bright sky and neither the computer nor the driver applied the brakes.

2. Reportedly, the city's plethora of bridges are the trickiest for the system to navigate.

3. In a society that has the ability to more than adequately provide sustenance and shelter for its people, most folks could spend significantly more time engaged in recreational, cultural, or entrepreneurial pursuits
but that would require the gains of industrialization and automation to be more widely distributed, rather than concentrated narrowly in the hands of capital owners.

1 comment:

Andrew T said...

a) Shoot, now I need to watch Minority Report again.

b) I'm a huge fan of the future of self-driving cars (get 'em to mass market within 5 years or so, I say!), but can't agree that they would remove the desire for personal ownership or start a trend of efficiently scheduled shared use. Ultimately cars are tools to serve schedules; the schedules don't revolve around the cars. Rush hour traffic doesn't exist because of some limitation on driving technology. In fact, companies and employees could already save themselves a lot of hassle by offering more flexible work hours to avoid driving at heavy times. Yet they don't because there are even greater incentives to keep piling all their driving demand on at the same time. First of all, people like to sleep during the night, so unless they're on an 8 hour trip*, those cars are going to need parking until dawn. But mostly, companies want the efficiency of people working at the same time, and families want the consistency of regular school hours with workdays to match. (While bachelors such as I are left to suffer the tyranny of the corporate/domestic masses.)

(*Admittedly, it's gonna be way cool to be able to go to sleep in DC and wake up in Boston!)

So the need for most vehicles inevitably is stuck in a pulsing cycle of peaks where many cars (autonomous or not) are needed and troughs where they must be parked. Sure, autonomy might help around the edges a bit, but the Rush Hour Problem won't really be solved until the REAL transportation tech breakthrough: robot Jackie Chan.

Wait, sorry, I meant perfect telepresence removing the need to drive there at all.

Still, looking forward to it!