Newsgroups: comp.ai.alife
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newsfeed.internetmci.com!csn!news-1.csn.net!gw1.att.com!gw2.att.com!oucsboss!scrapyard.ent.ohiou.edu!user
From: bushbo@bobcat.ent.ohiou.edu (Brian O. Bush)
Subject: Q: Robotics, the way nature intended
X-Nntp-Posting-Host: scrapyard.ent.ohiou.edu
Message-ID: <bushbo-1004960012280001@scrapyard.ent.ohiou.edu>
Keywords: robotics, alife, decentralized control, evolution
Sender: postmaster@scrapyard.ent.ohiou.edu
X-Nntp-Posting-Date: Wed Apr 10 00:12:26 1996
Organization: Ohio University
Date: Wed, 10 Apr 1996 04:12:28 GMT
Lines: 40

Hello.

I would like to start a discussion on the state of robotics. After just
glancing at the comp.robotics group, I am still somehow frustrated that
there is never much discussion on research, building or throwing ideas
around about more realistic robotics (not necessarily to solve a problem).
In this sense i am refering to the following:

    Not a centralized processor, instead many processors are utilized to
perform low level functions, such as reflex.
    Not focused on emulating a human, but rather simpler structures and
creatures
    Not focused on solving a particular problem, but more on how to
design a machine and the surrounding framework in order to apply
adaptation principles (evolutionary computation here?)

I know of the work done at MIT's mobot lab, Mark Tilden's BEAM robotics
principles. Both are at extremes on this machine intelligents, for Brook's
has cockroaches with complicated software structures, and Tilden has leg's
twitching when a capacitor reaches its saturation point. I was thinking
more on the lines between these two extremes. 

What comes to mind is a collection of robots, called the "herd", i saw one
time in Wired magazine and in the extinct AI Expert (both about a year or
so ago); these robots possessed a simple flocking behavior, as well as
other behaviors and were built by an artist and mechanical engineer (in
CA?). I never heard much about it after that,... but anyhow the robots
were built cheaply (main structure was grapevines) to demonstrate
technological 'life' as a piece of artwork. Not to control a satellite
from the space shuttle.

I guess many people would not want to build robots that are inherently out
of control and capable of adaptation. I was just wondering, for building
complex adaptive systems in software seems too easy (kinda like cheating
nature). Tell me your thoughts.

Cheers
Brian
-- 
Follow the Black Valley Trail of Death, Into the Beautiful Sunshine.
