Alt-BEAM Archive
Message #04969
To: beam@sgiblab.sgi.com
From: Ivar Thorson ivar@flashmail.com
Date: Wed, 30 Jun 1999 14:51:46 -0700
Subject: [alt-beam] Re: Bicore Feedback Experiments
Zozzles T. Freep wrote:
>> I think that seems to be an effective use of feedback, if you ask
me. It gets a little stuck,
>> and it shakes faster. It gets reallystuck, and it thrashes
violently with long strides until it is free.
>
> Ok, fine for the suspended bicore case. However, the "implex is a
good thing for a robot"
> was stated long before the community began using bicores on a
regular basis & I'm pretty
> sure that suspended bicores are (as Wilf originally stated) Really
Different than the 'everyday'
> nervous net.
And I totally agree with him. They are Really Different. I am just
experimenting with how they reacted to feedback.
> Nonetheless, if the leg is REALLY stuck, mightn't the thrashing
cause the robot to
> shake itself to pieces??? If I were to overextend the metaphor,
it reminds me of
> the hunter finding nothing left in his trap but the animal's
leg...
I don't know about you, but none of my robots have ever died my
shaking themselves to pieces (they usually die when I drop them.).
More often, the robot gets stuck in some stuff for several minutes
and would be able to escape if it moved its legs just a fraction of
a inch farther each way. In this case, the strange feedback
tendancies of the suspended bicore would indeed seem to be useful,
although I have not tested it and make no claims that it actually
works as advertised. ;-> As you said, it would be necessary to study
such bots for several weeks, although even then it would be
difficult to ascertain if the feedback was indeed useful to the
robot.
> When considering the case where the outputs are buffered, I've got
to ask the
> question I asked before: are you using the same chip for the
bicore & the
> buffering inverters?
Yes.
> (SNIP) particularly when you use a -separate- power supply for the
buffer/driver
> that you'd see different results.
I don't know. Possibly, although what exactly would happen is beyond
me.
> I also have to ask another question: did you read Wilf's report on
his experiment
> with regard to the feedback resistor?
Umm, I don't remember. Did it suggest that it was useless?
> In any case, a suspended bicore simply isn't the same animal as a
six Nv neuron
> loop, so the conclusions about the results/value
> of implex for one needn't (and probably don't) apply to the other.
Naturally. I did the same experiments with a Nv net, and the results
were markedly different.
> Ah well, only experimentation will tell.
Is that an offer? ;->
Ivar Thorson
------------------------------------------------------------------------
eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com
- Simplifying group communications
Home