December 21st, 2008


Basil and the Gundersons. Advanced robotics in Denver.

Then, with the work finished this past June, they wheeled a chair in front of Basil and asked him what he saw. Using the chipper dialogue they'd programmed, he announced, "Ooh, I see a wooden chair."

The regulars that night at the Denver Press Club, the Gundersons' favorite bar, didn't know what to make of the two scientists throwing down beers and dancing around chanting, "The robot saw a wooden chair!" But that's because they had no idea what this development meant: Reification worked. And with that, this tiny robotics lab started making waves.

"The reification work they have done is unique. There hasn't been a book on the subject anywhere else, and I consider them leaders in their field," says Raj Madhavan, a researcher at the Intelligent Systems Division at the National Institute of Standards & Technology in Gaithersburg, Maryland. "And some of the things they have been doing, the reification work, hadn't been implemented on a robot, per se. They have what looks like a toy robot, but it has a cybernetic brain design."

Soon Basil wasn't just identifying wooden chairs, but other types of chairs — followed by tables and even people. Now when the Gundersons show him a chair he's never seen before, he's can figure out that it's probably a chair and definitely not, say, a person or a table. Once the couple repairs Basil's little mobility problem, they expect him to be able to approach an unknown object, mosey around, scrutinize it for a bit, then place a basic mental model of the object in its memory for later use.

Basil, in other words, will be able to learn on his own.