Showing posts with label multi-agent systems. Show all posts
Showing posts with label multi-agent systems. Show all posts

Thursday, April 26, 2007

It Takes a Village…?

An interesting experiment in Scotland seeks to build a robotic society to study emergent cultural behaviors. The robots will be grouped into “villages” and programmed to mimic each other’s actions under various conditions. The scientists expect the mimicked behaviors to change slightly every time they are copied, resulting in a variety of different and unpredictable results...much like the telephone game. Of course, the emergent behaviors will be uniquely robotic, and should be distinct from human or animal activities.

Given the general sentiment towards AI in Great Britain lately, don’t be surprised to see protesters warning against the inevitability of these robots going crazy and killing everyone...

Wednesday, April 18, 2007

Ashes to Ashes...

Scottish engineers (alas, not Montgomery Scott) are speculating about tiny robotic "dust particles" which utilize wireless distributed computing and are capable of physically reconfiguring themselves. Consisting of multitudes of individual computer chips surrounded by morphic polymer sheaths, clouds of smart dust could fly in formation and changing shape to meet environmental and mission requirements. These devices would form a swarm intelligence, and could be used to explore alien worlds if released into the atmosphere of extrasolar planets by interplanetary space probes.

Talk of simulated swarms of 50 such particles able to shift into star formation immediately reminded me of one futurist's concept of a starfish-like body built of nanorobots designed for sophisticated AI beings of the future. Absolutely wild.