The future of robots

Just what is a robot?

Mention the word, and most people think of C-3PO from Star Wars or Data from Star Trek—a mechanical contrivance that acts like a human being. Actually, a robot is any self-directed machine that does our work. The term itself comes from the Czech word “robota” (“drudgery”) and was coined by Josef Capek for his brother Karel’s 1921 drama R.U.R., or Rossum’s Universal Robots. Capek depicts a world where robots perform all manual labor—until they rebel and destroy their human masters.

Who first conceived the idea?

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

Leonardo da Vinci, naturally. He first drew up plans for a mechanical man in 1495. But the idea wasn’t realized until the 17th and 18th centuries, when inventors created automatons with self-contained clockwork drives as playthings for European aristocrats. One of the most famous, built in 1774 by the Swiss inventors Pierre and Henri-Louis Jaquet Droz, was the Automatic Scribe—a lifelike boy that could write messages provided by its operator. The first truly functional robot, though, was created in the U.S. by Joseph Engelberger and George Devol. In 1961 they installed Unimate (short for Universal Automation), a programmable 4,000-pound mechanical arm, on a GM assembly line in New Jersey.

But what about robots that think?

That breakthrough came in 1968, when the Stanford Research Institute unveiled Shakey, the first mobile robot to use artificial intelligence to control its actions. Equipped with a primitive problem-solving program, Shakey used a TV camera, laser ranging, and bump sensors to navigate its way around objects. However, it took half an hour to move one meter.

How far have we come since then?

A long way, thanks to advances in areas like microminiaturization and infrared range finding. Some of the robots devised by a new generation of technicians, mainly in Japan, are truly amazing. Using a voice- and face-recognition system to take orders, Honda’s 4-foot-tall Asimo can walk up and down stairs, turn on lights, open a bottle of water, and fetch the mail. With sensors in its feet, Sony’s pint-size Qrio can run, stand on one foot, and get up after a fall. Robovie IIS, invented at Japan’s Advanced Telecommunications Research Institute International, is touch sensitive. Tap it on the shoulder, and it will turn and ask, “Yes?” If poked or hit, it will exclaim, “Ow!” But for all their innovation, these devices are still quite crude.

Why aren’t they more advanced?

The biggest barrier is creating true artificial intelligence. Human brains can carry out about 10,000 trillion operations per second. By contrast, the world’s most advanced supercomputer—IBM’s ASCI Purple—can perform only 100 trillion. Scientists have learned that even the simplest tasks, such as tying shoes, require an astonishing amount of sensory analysis, computation, and response to feedback. “Most robotic systems don’t actively learn from their experiences,” says Reid Simmons of Carnegie Mellon University’s Robotics Institute. “That would make a big difference if we could crack that nut. But right now everything is programmed in.”

Are there other barriers?

Yes: getting robots to move fluidly. Currently, robotic limbs require large power supplies and complex computer calculations to mimic a stiff version of human movement. Scientists think robots would move more efficiently if they had artificial muscles that worked like ours. But these muscles would have to be built from materials that are thousands of times stronger, lighter, and more flexible than anything we have now.

So what’s the best we can expect?

For the immediate future, experts say the most sophisticated robots will continue to be single-function machines like the self-navigating Roomba vacuum cleaner and Sweden’s Husqvarna self-propelled lawn mower. But they also predict that there will be a growing number of immobile robots, or “immobots.” These are household devices that incorporate software programs to automate a variety of their functions. The TiVo digital video recorder is a current example, and so is the programmable coffee maker that can grind beans and brew them while you’re still asleep. But the real robot explosion is not likely to occur in our living rooms.

Where will it occur?

In factories. Today, 90 percent of the world’s robots are in factories, with 137,000 of them in the U.S. alone. There they weld auto bodies, lift crates, solder wires and semiconductors, and perform other repetitive or back-breaking labor. It’s a far cry from Forbidden Planet’s Robby the Robot, acknowledges David Akin, a University of Maryland robotics expert: “We are at least a couple of miracles short of the full goal of robotics: an android that is indistinguishable from a human.” Some experts think that these miracles may be decades or even a century away. Others say a major breakthrough in artificial-intelligence research could come at any time—making humanoid robots a reality. “The question is, What is this magic juice that’s missing?” asks Jonathan Connell, an IBM researcher. “Once we understand what humanlike thought is, we’ll be able to make it.”

When Fido is a machine