The Robot That Does A Simple Job Very Well May Be Wave of Future

by Lee Gomes

With the movie based on the sci-fi novel “I, Robot” opening in a few weeks, there will be a lot of walking, talking robots on the screen this summer.p.

They’ll be Hollywood fiction, however. For real life, go to the Hines VA Hospital in Chicago and watch as a robotic wheelchair is ordered to roll out of one room, down a hall and into another room. The wheelchair can accomplish the task, but the process is painfully slow and methodical. The journey of a few dozen feet takes several minutes.p.

That hesitating performance would hardly impress most lay observers as state-of-the-art robotics. But it is. Says Steven B. Skaar, the wheelchair’s creator, “It’s hard to believe, but I don’t think there is another robot in the world that can do what this one is doing.”p.

Prof. Skaar teaches robotics at the University of Notre Dame . He is also something of an iconoclast within his field, a fact for which he may be paying a steep price.p.

Dr. Skaar is disdainful of much of university robotics research in the U.S., believing it exists in a kind of Emperor’s New Clothes world where academicians, always in search of grant money, won’t admit to themselves or others how little progress they are actually making.p.

Robotics, he says, is strewn with grand projects that never delivered on their original promises. He can recite, with a bit of schadenfreude, a kind of robotics Hall of Shame, like the current difficulty NASA is having in finding robots to repair the Hubble space telescope.p.

Actually, Dr. Skaar’s robotic wheelchair is something of an antirobot. The chair has no “intelligence” of its own. It can roll only along a path it has previously been “taught,” while using a huge amount of computing power to know where it is at any given time.p.

That’s an approach to robotics called “teach and repeat,” and it is commonly used for “dumb” factory robots. The approach is ordinarily not considered intellectually interesting by researchers, who are eager to cram intelligence into their machines. But Dr. Skaar says that interesting or not, it may be the best way to deliver on the original promise of robots.p.

Even a robot optimist who disagreed with Dr. Skaar would be forced to concede that the world’s robot population today is vastly smaller than was predicted a generation ago. Industrial interest in robots, for instance, has cooled. You may see lots of robotic arms in car commercials, but there are far fewer of them in real life than you might imagine. Honda’s famous Asimo walking robot is little more than a balancing machine; the robot can’t do anything but walk.p.

Why? Because of many of the same problems that have plagued the sister field of artificial intelligence. The big surprise of robotics research has been that even simple physical activities, like reaching out to grab an object, are as deeply complex as “advanced” human traits like language and logic.p.

Dr. Skaar is 51 years old, and he has been at Notre Dame since 1989. He came to robotics from another field — aerospace — and was unencumbered by any robotic orthodoxies. He knew how to control a satellite, and he decided to try to control a robotic arm the same way.p.

It’s a seemingly small, but quite important, problem in robotics that didn’t previously have a great solution. The system, which Dr. Skaar calls “camera-space manipulation,” works splendidly. It is being used to stack pallets of paper bags at Chicago’s Smurfit Stone Container Corp. Clearly, Dr. Skaar doesn’t just complain about lack of progress in his field but instead tries to do something about it.p.

It’s the sort of accomplishment that would have made others a celebrity in the field. It hasn’t happened to Dr. Skaar, though. For instance, he currently oversees a measly $100,000 in grants, something that’s a source of tension with campus higher-ups, who would like him to be bringing home more R&D bacon.p.

Why the cool reaction? Maybe he’s too much the skeptic. Maybe it’s because industry has heard too much about robotics breakthroughs and has stopped listening. Or maybe it’s because of simple prejudices. If “camera space manipulation” is indeed a great idea, then it surely would have come from one of the country’s elite research universities and not a campus best known for football.p.

Dr. Skaar admits to being despondent now and then about the possibility of ending up as what he calls an “also-ran.” He also frets that his relative isolation may make it difficult for his graduate students to get jobs. (That latter prospect would indeed be a pity. Guillermo Del Castillo, a brand-new Notre Dame Ph.D. who did much of the work on the wheelchair, has every bit of the brilliance of counterparts from the country’s best-known computer schools — but none of their usual cockiness.)p.

The professor, though, says he cheers himself up with C.S. Lewis’s admonition to pursue not the “inner ring” of peer approval but rather a craftsman’s sense of satisfaction.p.

And he presses on with his work. Maybe someone should make a movie about that.p.

TopicID: 5845