Newsgroups: comp.robotics.misc,sci.space.policy
Path: cantaloupe.srv.cs.cmu.edu!honeysuckle!das-news2.harvard.edu!oitnews.harvard.edu!news.cs.umb.edu!ulowell.uml.edu!cam-news-feed2.bbnplanet.com!news.bbnplanet.com!news.enteract.com!newsfeed.enteract.com!news.maxwell.syr.edu!ix.netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: Nonsense from JPL
Message-ID: <nagleEEwwAt.1wB@netcom.com>
Organization: Netcom On-Line Services
X-Newsreader: NN version 6.5.0 CURRENT #9
References: <871264321snz@stellar.demon.co.uk> <33EF5A14.74A3@schools.tec.mn.us> <871416808snz@stellar.demon.co.uk> <erkyrathEEu1n4.5FM@netcom.com> <33F2AF42.45D0@slip.net>
Date: Thu, 14 Aug 1997 16:25:41 GMT
Lines: 35
Sender: nagle@netcom10.netcom.com
Xref: glinda.oz.cs.cmu.edu comp.robotics.misc:16497 sci.space.policy:77028

Randy Gardner <randyg@slip.net> writes:
>Until we create far more sucessfull AI programs, the programming would
>be impossible.

>The problem is that each and every leg, arm, etc... would have to be
>programmed.  Our current programs can't think "How would I get from
>point A to point B?" and then do something like creating tentacles to
>cross ravines, etc....  With our current programs, to create a leg, we
>would have to write a seperate sub to create a leg, and one to move a
>leg, and one to lift a leg, etc.... Even after we do all of that, let's
>say we want it to walk differently, we would have to write a completeley
>different set of subs to do that, and with the almost infinate number of
>ways for a fractal robot to walk, this makes the descriptions
>impossible. We just can't write programs that can understand how
>movement works, and can create new ways itself, to do this would need a
>self-aware program.

     Programming something like that for legged locomotion isn't all
that hard.  More complicated problems than that have been solved.
See Mark Yin's PhD thesis at Stanford.  Read IEEE Trans. on Robotics,
or Latoumbe's book.

     It's the mechanics that's the problem.  You have to have a
large number of modules, each with its own power and drive,
and for most moves, most of them are dead weight.  From the 
pictures, most of the time, a single cube is providing the power
for a move; they don't combine their power much, if at all.
(That's a downside of non-backdrivable worm drives; you can't run
them in parallel because they jam.  Worm drives can be backdrivable,
though, if you use high-angle leadscrews.  See the Berg catalog.)

     Remember the movie "Demon Seed"?  It had a machine something
like this, but it used rotational joints.

						John Nagle
