May 17, 2005

"Self-replicating" Robots, Ha.

Go Forth and Multiply, Little Bot

Lots of groups have been reporting on self-replicating robots... which I'm finding a little tedious. If you take a look at the Wired link, it's not exactly what they're reporting....

The robot has to be fed -- receive new modules -- at just the right location and time in order to reproduce itself. To the biologically minded, that might not sound like true reproduction, but Lipson says all self-replication is dependent on the correct environmental conditions. Without food, animals can't reproduce. Without new modules, the Cornell bot can't make new versions of itself.
But, it's not really self-replicating then, is it?

Self-replicating means it can, given the right unfettered conditions, reproduce itself without material help from external sources -- or, at least, that's what I believe. That's not what you have in this situation. The modules which make up the robot are placed at "feeding stations", to be picked up by the robot. If you dropped these robots in the wild, it could never reproduce itself.

Having said that, I can see how it really *is* amazing on three points. First, it's design is very flexible and intelligent, allowing for fluid movement and close-quarters directional changes. Second, it's magnetic attachment is extremely cool, in that it allows for parts to be added dynamically to the robot, and those parts become part of the whole. That alone is worth a mention. And finally, it's made up of duplicates of a single part, which act in concert with one another. All of these are really difficult robotics tasks which are completely unique.

Just don't call it "self-replicating".

Posted by Ted Stevko at May 17, 2005 12:47 AM | TrackBack