LiveScience Research with a Stamp

bobhillier

I noticed a LiveScience article called·Eek! Robo-Rats Make Their Own Rules· at http://www.livescience.com/technology/050216_robo_rat.html. The robot definately looks like a Basic Stamp. I found it interesting that such simplistic functionality is recognized as learning to adapt and demonstrate nee, not-programmed behaviour.



▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
······ Bob, Ottawa
W75 54 17·· N45 18 30
·········· G16 #27

Jon Williams

Thanks for the link -- that is actually our Javelin Stamp processor.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax
Dallas, TX· USA

Paul Baker

It is AI but very primitive, as an undergrad I sat in on a seminar by an AI research team for another university (sorry can't remember which one). They created neuro-muscleskeletal models of fish and provided it a neural net brain roughly equivalent in processing power to a goldfish's brain. The brain was only given the the following perameter: propulse your body to acheive forward motion expending the minimal amount of energy. The AI brain at first convulsed the body in an unpredictable fashion, but in a suprising short period of time, the AI model of the goldfish was swimming identically to a real goldfish. At the time of the seminar they were undergoing predetor/prey modeling, where many AI goldfish were placed in a virtual tank with pillars as obstacles and a AI shark. After all models taught themselves how to swim in isolation, they were placed in the tank together. The AI shark was given the parameter: you must eat (intercept a AI goldfish) and the AI goldfish (there were about 50 or so in the tank) were given the parameter: you must not be eaten (avoid the shark AI). Quickly the goldfish learned schooling behavior with several learning leader behavior and others learning follower behavior. The shark quickly learned to pick out strays and outlyers of the school. Most facinating of all, some goldfish, while in schooling behavior and the school skirted around the pillars would dodge around the pillar keeping the pillar between themselves and the shark, rejoining the school when they felt it was safe to do so. At the time they were nearly finished with the fish models and were contemplating doing dinosaurs next. I've always wondered how their experiements progressed after the lecture, but I've been unable to find discussions of this work online.

Paul