Alt-BEAM Archive

Message #05217



To: beam@sgiblab.sgi.com
From: "Jean auBois" aubois@trail.com
Date: Sat, 10 Jul 1999 00:46:35 -0600
Subject: [alt-beam] Smart bodies???


Oh well, this is a contentious statement, but I think a gross exaggeration
is being continually repeated and it kind of gets in the way.

Current BEAM / nervous net robot bodies are simply not very smart. Again,
until there are some fairly "smart" BEAM / nervous net robots then a lot of
the argument about whether one starts with wild robots and then tame them or
starts with known devices and make them living machines is pretty much
academic.

By that way, there are no smart BEAM / nervous net robot limbs (although
between Spyder and Lampbot there is some potential.)

Mostly what we've got is something on the order of a set of perhaps two to
four somewhat smart joints... being controlled by something that isn't very
smart with the possible exception of the Unicore. And Unicore design
doesn't have a useful set of design rules yet, does it, so just how do you
figure out how smart it is???

Oh yeah -- so that I don't stupidly start another flame war like I did
earlier this week (I apologize!), how about I -declare- what "smart" means?
For the purposes of this conversation, "smart" -only- means the number of
individual, nameable things that a thing can do/perform. Obviously, my
definition really sucks because it doesn't take into the concept of
intelligence or anything else like that, but it does provide a fairly
objective way of judging things for the purposes of comparison. For example
(using sadly inadequate lists) let's compare an ant with the Walkman:

ant
role changes as it gets older
several tending behaviors related to reproduction of the nest (eggs,
larvae, pupae feeding & moving and all that)
dead ant removal
structure building including temperature and humidity control
active, definitely reactive defense and occasionally offense
several behaviors related to food (collection, processing --
particularly impressive with leafcutters, distribution, etc.)
actual reproduction of more ants (keeps them good old Selfish Genes
(Richard Dawkins) going, eh?)
movement, including overcoming obstacles
social and reactive navigation (pheromones vs. solar polarization or
whatever they use)
tending and control of other insects (for example, aphids)
etc.

Walkman
reactive defense
accidental offense & pushing things about
some behaviors related to "food" (locating best source, energy
conversion)
movement, including overcoming obstacles.
etc.

It would no doubt take a long time to get lists that are anywhere near
complete, but hey it looks like the ant's list is going to be at -least-
twice as long as the Walkman's... and a lot of it has to do with the
complexity ("smartness"?) differences in body design & whatever underlying
"infrastructure" controls it.


Now, I repeat that my definition of "smart" is pretty horrid, but I'll bet
you could rate solarengines, symets, solarrollers, photopoppers, two motor
walkers, four motor walkers, Stryder/Roswells, walkers with waists, Unicore
machines, Spyders using it and the result would be pretty much as you expect
it (that is, the first one in this list is the "least smart" and the last
ones are the "most smart".) So do we have anything even CLOSE to "smart"
bodies yet?


Have a good weekend!

Zoz


------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com - Simplifying group communications



Home