Biological Models of Computing and Self-Reconfiguring Robots


I'm attending the Biological Models of Computing session tonight. John Scott is the moderator. When I was at the University of Idaho, I was in a weekly seminar with two other CS faculty members and two faculty members from the Biology department where we read books and papers to explore this topic. I've been out of it for a long time, but it sounded like an interesting way to spend an evening.

We just played a game that shows emergent behavior. In the game, everyone picks two people at random and identifies one as the aggressor and one as the defender and when the moderator says "go" you try to position yourself so that you get behind your defender from the point of view of the aggressor. Then you make a simple change where you are the defender and try to position yourself between the aggressor and your other pick. Two very different kinds of behavior emerge. I won't say how it turns out so as to not ruin the game for you. Try it. The behavior is predictable if you've seen the game before, but not easy to predict without empirical data.

John works for Icosystem. The question of the day, because of the background of Icosystem is "what coherent behavior emerges from lots of independent actors each following their own set of rules. There are many other "biology inspired" computing models (viz. neural nets, genetic algorithms).

The USC CONRO project was mentioned in passing, so I looked it up. Very cool, self-reconfiguring robots. Check out the hardware prototypes and the movies on the software hormones page. The concept appears to be similar to what Rodney Brooks has been talking about, but in a cellular construction. These things are like Legos on steroids. Imagine being able to create 1000s of these kinds of things with self-contained power and good sensory equipment on each one. Who couldn't love playing with these?