In warehouses, hospitals and retail stores, and on city streets, industrial parks and the footpaths of college campuses, the first representatives of this new invading force are starting to become apparent.
“The robots are among us,” says Steve Jurvetson, a Silicon Valley investor and a director at Elon Musk’s Tesla and SpaceX companies, which have relied heavily on robotics. A multitude of machines will follow, he says: “A lot of people are going to come in contact with robots in the next two to five years.”
The arrival of the robots—and their potentially devastating effect on human employment—has been widely predicted. Now, the machines are starting to roll or walk out of the labs. In the process, they are about to tip off a financing boom as robotics—and artificial intelligence—becomes one of the hottest new markets in tech.
Technological advances mean that robots are increasingly looking and behaving like humans. But can these machines ever match their human counterparts? Financial Times's Richard Waters and Kana Inagaki discover the latest humanoid robots and ask whether these machines can truly enhance society.
After growing at a compound rate of 17 percent a year, the robot market will be worth $135 billion by 2019, according to IDC, a tech research firm. A boom is taking place in Asia, with Japan and China, which is in the early stages of retooling its manufacturing sector, accounting for 69 percent of all robot spending.
Although the amount of money flowing into a new robotics industry is still at a relatively early stage, all the lead indicators of the innovation economy are pointing up. Patent filings covering robotics technology—one sign of the expected impact—have soared. According to IFI Claims, a patent research company, annual filings have tripled over the past decade. China alone accounted for 35 percent of robot-related patent filings last year—more than double nearest rival Japan.
In another sign of the expected boom, venture capital investments more than doubled last year to $587 million, according to research firm CB Insights.
A lot of people are going to come in contact with robots in the next two to five years.
Steve Jurvetson | Tesla, SpaceX
Other investors are also piling in, says Manish Kothari of SRI International, a Silicon Valley research and development lab that has spun off robot companies. From private equity investors looking to build portfolios of robot investments, to new “incubators” such as Playground, started by former Google robotics chief Andy Rubin, the investment options have been proliferating rapidly.
But in many cases, the amounts being invested still seem disarmingly modest. Like other disruptive technologies, the seeds of this revolution can be seen in start-ups that operate on a shoestring but have grandiose aims.
They include companies such as Dispatch, a Silicon Valley company that is testing an autonomous delivery vehicle—a smart-box on wheels—on two college campuses in the U.S. The start-up has raised only $2 million but is riding the wave in collapsing costs of sensors and advances in artificial intelligence that are making autonomous machines a reality.
“There is an exponential pace of improvement in hardware and machine learning algorithms,” says co-founder Uriah Baalke. “The computational power required has gone down a lot.” The result is a new class of machines that can operate by themselves in human space, the advance guard of a new robot industry.
Until now, most robots have taken the form of expensive, high-precision industrial machines. Usually found operating in protective cages on automobile assembly lines, they have carried out preprogrammed tasks, with no need or scope to adapt to changing conditions.
Roboticist Hiroshi Ishiguro of Osaka University built his own mechanical twin to see how humans react to extraordinarily lifelike machines. [Photograph By Max Aguilera-Hellweg, National Geographic Creative]
The cheaper, flexible machines that are emerging are designed to be more adaptive. From driverless cars and drones to the “cobots” that work alongside humans in industrial settings, they try to sense and adapt to their surroundings. Like Tug (a robot that moves supplies around hospitals), Savioke (which handles deliveries to hotel rooms) and Locus Robotics (which operates in warehouses), they are moving into the service industries. In industrial settings—still the main venue for robot investment—they are moving out of the cages and into a far wider range of roles.
Like the arrival of PCs, the new era promises to take the technology into many more areas of working life. “The traditional industrial robots are mainframes—what we’re doing are PCs,” says Scott Eckert, chief executive of Rethink Robotics, a U.S. company whose robots help with packing or tend machines. Rethink says that the all-in cost of its Sawyer robotic arm amounts to about $1 an hour, a price at which many of the jobs that have been beyond the reach of automation could be affected.
The technology advances behind this wave of innovation have come together remarkably quickly. Funding over the past five years by Darpa, the research arm of the U.S. Defense Department, has brought breakthroughs in mechanical areas such as robotic limbs, says SRI International’s Kothari.
But the biggest advances have come in software. Improvements in computer vision, for instance, have made possible many companies like Dispatch, whose machines rely on being able to “see” the world around them, says Chris Dixon, a partner at venture capital firm Andreessen Horowitz.
Machine learning algorithms, which are designed to adapt through an endless process of trial and error, play the biggest part in teaching robots how to navigate a world beyond the normal rules-based systems that computers are designed to handle.
“You won’t have to programmatically tell it what to do; it will figure it out,” says Vinod Khosla, a venture capitalist who has backed robot companies in markets including agriculture and health care. “Today, it’s really dumb intelligence—but that will change quickly.”
When it comes to designing the machines for this emerging industry, most robot entrepreneurs and investors are following a similar formula.
One element is to build low-cost machines that tackle specific tasks, rather than attempt to create general-purpose machines—let alone fully humanoid robots—that try to take on too much.
One of the world’s first “emotional” robots, Pepper visits the Financial Times where columnist and FT.com managing editor Robert Shrimsley attempts to make conversation. The robot's creators Softbank of Japan and French subsidiary Aldebaran claim Pepper can understand human emotions.
The goal is to build “single-purpose robots that do one thing very well”, says Dmitry Grishin, a Russian who recently raised a $100 million fund to invest in robots and other hardware. If they succeed, these machines quickly lose their status as “robots” and become more part of the fabric of everyday life, he says—like automated vacuum cleaners or cash machines.
Another design feature of many of the early robots is to operate alongside people, making humans more productive rather than replacing them altogether. Many of these robots, for instance, hand over decision-making to a human operator when they encounter situations they cannot understand or navigate.
“The truth is, anyone who works in robotics knows the limitations of what they’re working with, and they’re pretty extensive,” says Kothari. Robot companies also want to keep “the human in the loop” because they believe it will make their machines more socially acceptable and less threatening, he says. Most people operating in the robot industry say humans will have an important role to play in directing the machines for decades to come.
There isn’t a single mechanical or physical thing a human will be able to do better than a robot.
Steve Jurvetson | Tesla, SpaceX
That does not change the long-term threat to jobs, however. “There isn’t a single mechanical or physical thing a human will be able to do better than a robot,” says Tesla and SpaceX’s Jurvetson.
Another feature the robot makers are counting on is to be able to use the learning capabilities of their initial products to achieve rapid improvements and gain an advantage over rivals that are slower to get their machines into the market.
“Once you ship the device, you can apply more and more intelligence and machine learning,” says Grishin, the Russian robot investor. The trick, he says, will be to find a task that the relatively dumb machines are able to handle, then use knowledge gained in the field to rapidly add to their capabilities and usefulness. “First put them in consumers’ hands, then learn from their behavior.”
This is the secret weapon that all robot companies rely on. “Everything gets better over time,” says Jurvetson. “This is happening in almost every hardware product: they are becoming minimal vessels for software.”
This technological shift has set traditional robotics leaders in Japan and Germany against nascent industries in countries such as the U.S. and China.
“Right now, the U.S. is definitely the leader” when it comes to software, says Grishin. He adds, however, that the hardware manufacturing expertise of China makes that country a contender, particularly since robotics has become a national priority. As a result, the rise of a new robot industry is about to trigger a global race for leadership.