Robots . . . Robots Who Need People

If I remember the Popular Science and Mechanix Illustrated magazines of my youth correctly, by the end of the millennium we are supposed to relax in our easy chairs while robots mow our lawns, wash our windows and vacuum our rugs.

Well, the year 2000 is nigh upon us, and around my house, humans are still the only robots doing those chores.

But while I still have to mow my own lawn, there is a growing chance that if I ever need brain surgery, a robot will do the job.

Robots are increasingly important in the modern world, but not for the tasks that many of us expected.

The fundamental reason is not complexity, but cost. It makes sense for a hospital, or a major manufacturer, to pay millions for robots that will make their work more effective, but “no one at home is willing to pay $20,000 for a robot that does a small number of tasks well,” says Jean-Claude Latombe, chairman of the computer sciences department at Stanford University.

Stanford has long been at the forefront in robotics technology, and Latombe says he has seen a major shift in philosophy. Robotics began with an emphasis on mechanical devices that could function with virtually no human intervention, but it has evolved into computer-driven systems where software, not hardware, is the principal driver, and a human is often overseeing the process.

“One of the fastest-growing areas for robotics today is robot-assisted surgery,” Latombe says. “Here at Stanford, we are doing brain surgery with robots.”

The system has advanced to the point that the surgeon is not even in the same room as the patient during radiation surgery to remove brain tumors. Radiation in the room is too great for repeated exposure by the surgeon. So instead, the doctor sits at a console in a nearby room, monitoring the operation while the robot does its thing.

“The brain is a complex tissue where you have critical tissue that you absolutely don’t want to radiate because you can have very bad side effects, like making the person blind,” Latombe says. “And at the same time, you want to put a lot of radiation in the tumor because otherwise you will not treat the tumor.”

Even the slightest movement by the patient could have catastrophic results, so sensors detect every twitch and relay that to the robot, which makes automatic adjustments to offset the movement.

Robotic devices could eventually play key roles in many types of surgeries. A Sacramento firm, Integrated Surgical Systems, has developed a robot called “Robodoc” that bores precision holes in a patient’s thighbone during hip replacement surgery. The device is now being used in Europe and is awaiting approval by the U.S. Food and Drug Administration for use in this country.

Despite such technological breakthroughs in surgery, manufacturing still remains the primary user of robotics, although Latombe says even that got off to a bad start.

Like most of us, manufacturers initially expected too much of robots. They wanted them to be completely autonomous, virtually eliminating humans from the process. That didn’t work, Latombe says, because early robots were expected to be too versatile.

*

Newer robots are designed to do precise tasks, and they have caught on in a big way. Latombe says that a decade ago, when he toured auto-manufacturing plants, he saw “lots and lots of humans. Now they don’t have a lot of humans.”

One plant he studied recently has at least 100 robots, but humans are still directing the work. Far from the assembly-line workers of yesteryear, they are highly trained computer technicians.

Robots may have changed the workplace, but autonomous gadgets that will ease our burdens around the home are still hard to find.

Keith L. Doty, director of the Machine Intelligence Laboratory at the University of Florida, says he has been inundated with requests from potential customers ever since announcing that he and his students have built a robotic lawn mower that can cut your grass while you are at the beach.

Doty’s battery-operated LawnNibbler uses sonar and infrared sensors to know where it is at all times as it cuts a 6-inch swath across the grass. The sensors also warn if it is about to bump into a tree, or the family pooch has settled down in its path, and its computer guides it around the obstacle.

He says it represents more of an engineering triumph than a robotics breakthrough, and he thinks one of these days consumers might be able to pick one up at the local hardware store for about $1,000.

There are still a few bugs to work out of the system, he says, but “we see the light at the end of the tunnel. It’s not like there’s some critical problem that needs to be resolved.”

Robots that carry out basic chores are already in widespread use. Undersea robots developed at USC, for example, have gathered data from the bottom of harbors, seas and oceans all over Earth. And of course there is the Sojourner and the fantastic flying machines developed at the Jet Propulsion Laboratory in Pasadena to explore other planets.

Those robots find their strength in the fact that they operate in areas where it would be very difficult, even impossible, for humans to work.

At the Massachusetts Institute of Technology, often cited as the birthplace of artificial intelligence, no one seems worried about such mundane things as building robotic lawn mowers. They want to build a real robot that acts and works and to some degree even looks like a human being.

MIT’s celebrated robot, Cog, is a humanoid, meaning it looks as much like a human as they have been able to make it. But why should anyone care what a robot looks like if it can do the job for which it was built?

According to scientists at MIT, even a robot needs a pretty face. People will never be able to interact properly with a robot that doesn’t look like a human, according to researchers working on Cog. It doesn’t look like any human I’ve ever seen, but it does have arms and legs and a head.

“If the robot has humanoid form, then it will be both easy and natural for humans to interact with it in a human-like way,” according to one document that lays out the rationale behind the Cog design.

*

“In fact it has been our observation that with just a very few human-like cues from a humanoid robot, people naturally fall into the pattern of interacting with it as if it were a human.”

A “disembodied human intelligence,” according to the MIT dogma, won’t suffice, even if it does do windows.

Of course, that makes the task vastly more difficult. If we have to wait for robots that can taste, smell and watch Monday night football with us, we are probably going to have to wait a long, long time.

But just how much of this do we need?

Scientists at Silsoe Research Institute and Oxford University Computing Laboratory in England recently embarked on a three-year effort to build robots that can herd a flock of ducks. Stephen Cameron, one of the researchers, outlined the goals during an artificial-intelligence conference recently.

“We’re trying to produce a robot that reacts a little bit like a sheep dog,” says Cameron.

My border collie was really upset about that one.

*

Lee Dye can be reached via e-mail at [email protected]