There are now some 120,000 warehouses globally, and another 50,000 are likely to be added before 2025. Over the next few years, more robots will be deployed into these warehouses—the logistics market—than in all other application categories combined, including farming, medicine, and home use. Just as the 1960s saw the mechanization of industry, with an accompanying boom in productivity and prosperity, the 2020s will be the dawn of the robotification of services.
Industrial robots came into use in 1961 when General Motors (G.M.) installed a simple robotic arm on its New Jersey production line. The machine had been invented by Unimation, a company founded by the father of robotics, Joseph Engelberger—a self-professed Isaac Asimov enthusiast. By 1969, G.M. rebuilt its Lordstown, Ohio, factory with an array of Unimates to perform welds, and the facility soon achieved a twofold leap over its former production rate, making it the most productive factory in the world. (That same factory would be sold in 2020 to startup Lordstown Motors, with plans to make electric trucks.) Automobile manufacturers everywhere were among the first and fastest to embrace industrial robots.
The International Federation of Robotics, founded in 1987, issues an annual robot census. When 2020 began, it found nearly 400 million industrial robots at work in factories around the world, twice the number from five years earlier. But for the first time, over half of all robot purchases globally were in services, not industrial applications. And while growth in the latter is expected to continue, installations of service bots are expected to rise more than 200 percent in just a couple of years.
About half of all service robots are found in the logistics market, with “inspection” applications at about one-fifth. The military, an early and ongoing supporter of robotic technology, accounts for only a tiny fraction of the market. The rest is made up of everything from professional cleaning and fruit picking to delivering medications in hospitals.
The service-bot counterpart to the robotification of G.M.’s Lordstown factory came in 2012, when Amazon spent $775 million to buy Kiva Systems. Kiva had invented a clever self-propelled turtle-like robot that can scoot around warehouses carrying entire shelf-units of packages.
Firms like Amazon and Walmart need more than the information exchanged between buyers and sellers in cyberspace; they need the physical exchanges that occur in warehouses. That’s how the seamless experience of “one-click” shopping happens. Kiva-class service bots are the cloud’s hands and feet, directly and wirelessly controlled by the cloud in real-time.
In the past decade, annual net additions to warehouse square footage have increased 400 percent. That helps explain a nearly identical 400 percent increase during the past five years in service robot sales into the warehouse and logistics supply chain.
E-commerce has done more than increase the demand for warehouses; it has changed their function. Before, pallets of goods arriving at a warehouse were redistributed, often again on pallets, to local retailers, where staff would unpack and sort the goods onto shelves for display. One-click e-commerce has pushed the latter half of this process back upstream into warehouses, many of them multi-story structures bigger than football fields, where single packages (down to a tube of toothpaste or a single book) are grabbed, boxed, and delivered directly to the consumer’s doorstep.
As e-commerce pushes more and smaller warehouses toward the edge of supply chain networks, closer to consumers, service bots solve another problem. Since such edge facilities are necessarily located where real estate is more expensive, operators chase greater efficiency in using a building’s floorspace by packing things in more tightly. In these high-density workspaces, it’s far safer to use robots and automated systems. The pinnacle of density, a kind of Rubik’s Cube–like design for bins and packages, leaves no room for people.
Whether in the hyper-dense local warehouses or the hyper-scale remote warehouses, the 2020s will see the emergence of what we might term a warehouse-scale robot. Human beings will still be involved, especially at the front end and the output, but the storing, moving, and sorting of packages will be autonomous, just as the storing, moving, and sorting of data is automated in a warehouse-scale computer.
Package-handling service bots are part of a broader warehouse automation trend, both inside the buildings and for that last mile. As performance and adaptability improve and as costs decline, robotification will come to every segment of the services sector, from security, safety, and environmental monitoring and assessment to education, farming, general-purpose cleaning, and health care. After the logistics market, service bot deployments are growing fastest in medicine and agriculture.
John Froelich invented the first farm tractor in 1891, and he formed a company a few years later called the Waterloo Gasoline Traction Energy Company, named to differentiate it from the existing but far inferior steam-driven machines of the day. In 1918 he sold the company to a plow manufacturer called John Deere.
Almost 80 years after that, John Deere’s Precision Farming Group began work on a GPS-guided tractor. That came to fruition in 2002, when the company debuted a GPS-guided self-navigating farm vehicle—the machine that marks the dawn of agricultural cyber-physical systems. Ubiquitous self-driving cars may still be years away, but at least two-thirds of North American crop acreage already uses self-guided cyber-physical machines. In Australia, the figure is 90 percent.
Overall, agriculture accounts for a relatively small piece of the robot market—about $5 billion a year. But that’s about to change. The cloud’s A.I. logic engine marks an improvement comparable to the gasoline engine’s improvement over steam. Instead of huge, expensive tractors, we’ll also soon see swarm farm robots: machines a fraction of the size and cost, useful for hyper-precise fertilizing and weeding, enabling smaller boutique farms to compete with their industrial-scale counterparts. Meanwhile, the combination of A.I.-enabled vision systems (to see whether the fruit is ripe) and soft materials for grabbers will finally lead to fruit-picking robots. And just in time: Demographic trends point to both a rising labor gap in agriculture and rising global demand for food.
Dr. Bot, M.D.
Medical robots are still in their infancy, but they’re also already a $5 billion industry. Two decades ago, the Food and Drug Administration (FDA) approved Intuitive Surgical’s da Vinci robot for use during endoscopic surgery. In 2017 it approved Mazor Robotics’ spinal surgery robot, and in 2018 it gave Auris Health the go-ahead for a robot used in endoscopic and lung surgeries.
Many of these innovations entail an elastic use of the word robot. Machines like the da Vinci would be more precisely described as “tele-operated” or “robot-assisted”: A person remotely operates the machine, and the machine assists in precision. In other words, the machine is not autonomous. The same is true of “exoskeletons,” a class of Iron Man–style devices that help humans lift and move heavy objects, minimizing strain and enhancing strength. While unwieldy and impractical prototypes date back to the 1960s, only recently have the maturation of lightweight materials, superior power systems, sensors, and A.I.-enabled software controls allowed us to build useful exoskeletons.
Exoskeletons are now beginning to show in medical, manufacturing, and even construction markets. For example, Sarcos Robotics recently put its latest system into field trials partnering with Delta Airlines for baggage handling. Sarcos makes the reasonable claim that for an all-in cost equivalent to about $25 per hour, an exoskeleton can increase employee productivity by four- to eightfold in heavy-lifting tasks.
In June 2020 another pioneer, Ekso Bionics, received FDA clearance to market an exoskeleton that helps patients with brain injuries regain their ability to walk. Similarly, after retiring its walking Asimo robot in 2018, Honda started applying the technology to wearable exoskeletons for the elderly. As exoskeletons become less costly and more durable and comfortable, we can expect wheelchairs to become a thing of the past. Exoskeletons are on track to become a multibillion-dollar industry in the 2020s.
One benefit from the introduction of service bots will be improved safety for employees in high-risk occupations. Nine out of 10 of the most dangerous occupations are in construction, landscaping, farming, ranching, and fishing, all fields being transformed by the robotics revolution.
The most interesting question now is not how much existing machines will improve but what entirely new kinds of machines are on the verge of commercial viability.
The annual industrial robot census does not count consumer machines. Nearly 20 million bots were sold on the consumer market in 2020. These are—for now—mostly low-cost devices for relatively low-value applications: vacuuming, mowing lawns, toys, etc. They are more akin to automatic washing machines than the anthropomorphic robots that writers like Asimov imagined.
As with the automobile and the smartphone, the true robot is made possible when a whole suite of technologies matures. Henry Ford could not have built his great enterprise but for the confluence of the gasoline engine, petroleum refining, and the assembly line, none of which he or his company invented. Similarly, the iPhone could not have been built were it not for the maturation of three technologies, none of which Apple had anything to do with inventing: the silicon microprocessor, the pocket-sized TV screen, and the lithium battery.
For useful, untethered robots, the three enabling technologies now maturing are vision “chips,” synthetic “muscles,” and lithium batteries. Put those together with ubiquitous supercomputing power, and a revolution becomes possible.
The collapsing size and increasing capabilities of vision chips were propelled not by the aspirations of roboticists but by the consumer market for digital cameras embedded in smartphones, automotive engineers chasing chip-sized radar for better cruise control, and other unrelated applications. The solution that has eluded roboticists for years, the ability to mimic muscles, now emerges from materials sciences, with electrical, pneumatic, and polymer actuators that have the necessary efficiency, power, range of motion, durability, and (soon) self-repair. And the onboard power to animate it all? We owe that to the lithium chemistry developed at Exxon in the late 1970s.
At the growth rates now underway, professional service robots will become an increasingly common part of everyday work life for a rapidly increasing fraction of the populace. Until now, the service sector has been infamously immune to the kinds of machine-driven productivity gains seen in factories and farms—gains that invariably create new kinds of work and a widespread growth in wealth.
When cars were first introduced, it was clear that profoundly superior performance was inevitable, even though the utility of the first Packard in 1899 wasn’t much greater than a horse-drawn wagon. The path forward, and the velocity of growth and change, became obvious in 1919 with the introduction of the Model T. By the end of the 1920s, about 20 percent of the population owned cars. Along the way, hundreds of U.S. automakers sprung up, creating fortunes and entirely new domains of direct and indirect employment.
In the 2020s, as robot manufacturing matures, home-owners will first begin to purchase lower-cost versions of service robots to, say, help the home-bound elderly. At that point, a new class of machine will have been literally domesticated. And who knows what will happen then? In the words of Steffi Paepcke, a team leader at the Toyota Research Institute: “If the inventors of the automobile had asked people riding horses what they wanted, they would have answered that they just wanted a faster horse. It can be difficult to imagine a future that’s vastly different from the status quo.”
Not everyone has been enthralled with the rise of the robots. In 1961, with automobile manufacturing jobs declining even as output soared, President John F. Kennedy created an Office of Automation and Manpower to address, as he put it, “the major domestic challenge of the Sixties: to maintain full employment at a time when automation, of course, is replacing men.” Today, about 60 percent of the kinds of jobs that existed in that period no longer exist.
But if labor-saving technology were a net job destroyer, the unemployment rate would have been continually rising over all of modern history. It hasn’t. As some skills cease to be essential, different types of work emerge instead. The robotification of services promises to bring us the same things the mechanization of industry did: more business, more services, more wealth, and more well-being.