Robot predictions are always afoot, but now Amazon and Google have drowned out the usual futurist calls. Can these corporations succeed where so many researchers and startups have failed: to successfully re-imagine the role of robots in society? Jeff Bezos says that Amazon will deliver your package by drone within five years. It is tempting to dismiss his prediction as a perfectly-timed publicity stunt for cyber Monday. But drones, and many other robotic forms, really are coming to sidewalks, parks and yards -- open spaces that, for centuries, have been the domain of humans and dogs.
Robot progress has been throttled by cost and computing limits; it was too expensive to make new forms, and onboard computers were too heavy and too exotic for many years. But the 3D printing revolution, the maker movement and cellphone chip advances have overcome all these barriers. A start-up company can design a new robot, supply it with heavy-duty computer vision processing, GPS and high-quality cameras, and manufacture the resulting prototypes in a matter of weeks. This heralds a Cambrian explosion of robots in terms of diversity and quantity. We already have robots that outrun people, robots that fly through windows and robot cameras that bounce through a house. And, eventually, we will certainly have delivery robots.
But, flying drones face significant obstacles that make Bezos' prediction absurdly optimistic. First, there are the legal challenges. The FAA must come up with a plan for how thousands of airborne domestic drones will avoid interfering with our manned flight airspace, and this plan will likely require the drones to fly below five hundred feet. But these laws, in turn, will mean that drones must safely deal with low altitude obstacles: balloons, kite string, power lines, towers, bridge cabling and flagpoles. The sensing challenge will be significant, if not insurmountable, on a tiny, flying platform where weight-savings are crucial.
Then there is the "last meter" problem. When a delivery drone is close to its destination, the ensuing landing decisions are not just mathematical optimizations any longer; they are social challenges that require civic-minded common sense and neighborliness. Don't deliver the package on top of a sunbather. Don't put the package on the edge of a step where someone will trip. Don't land in small gravel, kicking up a stone that breaks the living room window. Solving these problems will require a complex understanding of social context and civility, and that is something even Siri has not yet accomplished -- just watch it conversing happily with a Furby.
It is a blessing that all the electronic spam we face is relegated to our screens. The borders of our smartphones, laptops and televisions serve as powerful prison bars that keep spam from breaking out, growing wheels, and occupying the physical world. In my recent book Robot Futures, I call the chapter on the uncontrolled release of robots into the wild "Robot Smog," because these machines will become a new bridge from the digital world to our physical world. They will invade our spaces and senses, buzzing in our ears and lurking in the air, with none of us quite sure what they are about to do.
Robotic pollution is just one concern. Robots will become more capable and more richly integrated in society, and this is where Google's recent announcements become relevant. Google has begun a major effort in robotics by purchasing at least seven companies to build a strong starting lineup. The talents these acquisitions bring are telling: these are companies devoted to understanding the electronics and mechanics of manipulation and mobility. They build the most advanced wheel drives and dexterous arms on the market, and these will be the crown jewels of any effort to create robots that can manipulate every object in a factory and navigate smoothly on an even floor. The most obvious first application for Google's robotics project will be the highly controlled environment of a factory or warehouse where the contingencies of misplaced children's toys, tasseled carpets and staircases have no impact.
Will Google be able to build newly capable robots that can invade low-cost manufacturing globally? Without a doubt. It will take time, and a great deal of investment, but the robots that Google and others eventually birth will only decrease in cost and increase in capability, year after year. But, this trend-line leads us to a much larger issue that is touched by both Amazon and Google's ventures: chronic underemployment. In Race Against the Machine Brynjolfsson and McAfee make a cogent argument about the way in which increasing manufacturing productivity is no longer a net win. We aren't innovating new products fast enough, or growing economies reliably enough, to offset the fact that fewer middle class people build our goods. The dystopian scenario we may face is a recovery that is not only jobless, but, in the long run, chronically job-sapping. For every semi-skilled job, there is arguably a path to redundancy through automation, artificial intelligence and robotics.
The corporate heavyweights, Amazon and Google, will innovate far more effectively, with deeper pockets, than anyone thus far, and so the rate at which real jobs will be economically replaceable by smart, mobile, dexterous robots that work tirelessly, is set to increase. There is no easy Industrial Revolution way out of this trend -- robots will always improve in their ability to manipulate objects, navigate about our world and make intelligent decisions.
Simply put, we cannot count on humans always being better than robots at any particular task. Google's robots may be most profitable in manufacturing initially, but robots will grow legs rather than wheels, and their legged progeny will be able to function in most places that we do. The service industry is just as threatened by our robot future as manufacturing, but simply on a longer timeline. Amazon drones may seem comic right now; I would say they are at least a decade away from realization. But the unchecked robot future before us is forecast to include heavy doses of robot smog, and requiring ever-shrinking pockets of human labor, which is a far more frightening, long-term prospect.