Alt-BEAM Archive

Message #03890



To: beam@corp.sgi.com
From: Jean auBois aubois@trail.com
Date: Fri, 28 May 1999 15:26:41 -0600
Subject: [alt-beam] Handling Snakes


At 01:40 PM 5/28/99 , Wilf Rigter wrote:
(I've done some editing)

>A dazzling display of Beam robological slight of hand you say?
>
>The effect of removing one coupling resistor of a HCT240 bicore will not
>result in loss of "control lock" but will certainly "de-tune" the slave
>bicore and in a walker would probably result in turning motion or tipping
>over. The effect of this "de-tuning" would be less apparent in a wiggling
>snakebot.

Well, I've got to admit that to begin with it is rather kind of tough to
determine if a snakebot is tuned or de-tuned or completely unsynchronized.
I'm not EVEN going to address the quantitative, but I'll make a stab at the
qualitative.

One of the characteristics of the intact snakebot that people most strongly
notice is how lifelike it seems in spite of being made (for the most part)
of metal and plastic. Some of this can probably be attributed to the
rather slow swing of the motors and the force with which those swings are
made. I think that more important than this is how the swing of each
segment follows (lags) its ''predecessor" in an organized, sinuous manner
and that this lag is affected by how much of a load you put on the motors.
In any case, the organized behavior coupled with a degree of interactivity
seems to unsettle people; also, it is obvious that the robot does a pretty
good job of tumbling on flat surfaces and can climb over stuff about as
tall as it is wide (when it is resting.) If you want something to wander
around terrain, touching off mines, this beast would no doubt do a good job
... for at least a while.

Ok, say we take out one coupling resistor all the way down the chain (which
isn't exactly the damage I'd expect from a mine, but let's ignore that.)
As a result, the bicores are relatively de-tuned in comparison to the
intact state -- but does that matter much? I think it depends on the
duration of the swing of a given segment compared to the duration of the
lag before the next segment starts moving. If the lag is relatively small
and doesn't vary too much, the robot will still operate in a fairly
organized manner. If the lag is comparable to the swing duration (and,
again, it doesn't vary too much) the robot will seem to open and close --
it might still seem lifelike, but it isn't going to work very well as a
tumbler. Somewhere in there, if the lag is the right value, the robot will
seem to do its snakey thing, but backwards (a result shown in the lamprey
experiments.)

Perhaps a more important question is this: does de-tuning more affect the
lag or how much the lag varies from cycle to cycle? If it varies enough,
the robot may well still seem lifelike (perhaps even more so than in its
intact condition) because it will go from being sinuous to
contracting/expanding to heaven knows what else and in any order one could
imagine -- fairly chaotically. It would almost certainly spend most of its
time thrashing in one spot and only a little time tumbling.

Say we take out ALL of the coupling resistors. Under those conditions,
you'd have the three bicores running essentially independently, each with
its own period. I think that you'd get a situation similar to the
preceding paragraph. If power-supply loading causes a secondary tuning
(phase-lock) effect, it would be interesting to see just how strong it
would be. After all, if it exists, it is a factor in ALL of the other
permutations: it is just that given no explicit coupling between the
bicores this might actually produce an observable effect. Still, I think
that you'd get relatively little organized behavior (with regard to the
mine-destroying purpose) but perhaps the snakebot might still feel pretty
lifelike.

However, the question of power-supply loading brings up an important
survivability issue. At present, if you destroy either the segment that
contains the three bicores or the section with the battery, chances are
you've got a non-functional robot. I suppose that if you zapped the
portion of the bicore section furthest out from the "center" that the
remaining two and two-thirds of the 'bot might continue doing their thing.
If you zapped the section with the battery, however (or the connection
between the battery & the bicores & motor drivers if any) you would then
have a Dead Snakebot.

Lampbot is at least a 50% improvement on this. The control circuitry for
each segment is right there with that segment's motor. As long as the head
(that is where the batteries are, looking at the photographs) isn't
destroyed (or the connection between the head and the rest of Lampbot) then
whatever is connected to the head ought to still work after a fashion.

There is room for further improvement with regard to survivability, though
-- it seems to me that each segment ought to be pretty much independent
with the exception of coupling and mechanical connections. One could still
have a modified power bus which could provide redundancy both for power
(perhaps a battery on a single segment could power the entire 'bot???) and
control-coupling if all the explicit coupling was destroyed. If all of the
segments were essentially identical, they could be mass-produced and
configured in arbitrary lengths. The only question then would be ensuring
that one segment could be set up as the head and another as the tail
(presuming you don't want the middle of the snakebot/lampbot to be the
'boss' .)

I guess that, yes, Mark uses a "dazzling display of Beam robological slight
of hand" as you say. After all, although it might be impressive that one
can take out a bunch of resistors and the robot "keeps on ticking", what
does that really mean? Is the resulting behavior still actually
organized/useful? Furthermore, is the destruction of such minor components
likely to be the kind of damage that occurs? What happens to Snakebot if
the damage ends up shorting everything on one of the bicore boards?

Still, Snakebot and Spyder and all those other robots are rather remarkable
feats of engineering. For all of my kvetching, I want people to know that
I'm impressed.



---------------------------------------------------------------
jab

Do you have a BEAM question? Try one of the following sites -- they really
help!
BEAM Robotics Tek FAQ http://people.ne.mediaone.net/bushbo/beam/FAQ.html
BEAM Mailing List Archive http://www.egroups.com/list/alt-beam/
BEAM Online http://www.beam-online.com/
Chiu-Yuan's BEAM Page http://www.geocities.com/SouthBeach/6897/
BEAM Heretics http://www.serve.com/heretics/

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

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



Home