Alt-BEAM Archive

Message #04765



To: beam-elite@webconn.com
From: "Mark W. Tilden" mwtilden@aerie.lanl.gov
Date: Wed, 17 Dec 1997 18:14:05 -0800
Subject: Re: flame-gate



> Tough questions, critics critiques of BEAM..
>
> 1. The notion that one couldn't use capacitors bigger than a
> certain value or else it would blow the output of an IC
> -- Tilden found that if one was using electrolytics and made
> certain that the positive lead pointed toward the output line
> that everything was just fine & is using capacitors that are
> far larger than he used to.

Zoz reported this, but I must warn everyone that this has resulted
in a good number of bot deaths due to fast-frequency charge-pump
values taking out the inverter inputs. Also the caps can develop
resistive holes in them which can offset your timing parameters, so
if you do do this, be sure you use caps which are rated at least 5
times higher than the voltage you're running your bot from.

Better yet, stick to monolithics. Less heartbreak. Strider died
from it.

> 2. The notion that there is definite and discrete feedback
> between a motor and the operating characteristics of the
> -particular- nervous neuron that was feeding it:

The driving neuron is not affected, it is the process that is now
being handled by the subsequent neuron. Processes temporal states
are very sensitive to voltage changes, but more so when a feedback
across the driver is applied. The point is there are three forms
of known feedback, power attenuation, direct inductive feedback,
and feed-forward limiting through sensors. Each has it's own
degree of influence on the microcore process phasing.

> 3. That particular bit of stuff (i.e. that the robot adaptively
> reacts to its environment) has been repeated over and over.
> Note adapting to ones enviroment is different than evolving.
> ADAPT - implies a modification according to changing
> circumstances. Change of actions, 'behaviors', etc.

My definition of adaption is the ability to adjust to new or
different conditions. Conditions autonomous systems must adapt to
range from internal perturbations, to local, to external, to
social-tactical. Some Nv architectures allow you to handle all
adaption levels. Open loop Nv systems can handle some. Pure
mechanics handle less, but it is possible to handle (as shown by
experiment) one or all of these with a combination of minimal
mechanics, electronics, Nv layers, or feedback options. However, I
don't know the limits of each. Is it possible to build a purely
clockwork cogno-adaptive creature with no electronics or sensors?
I suspect yes but it's hard to say for sure.

> 4. Anthropamoriphisms.., actually...they are 'life-pomorphisms'
> vs. human: Words like: behavior, learn, adapt,etc.
> parameciums - basically just sense for pH and move away from
> acidic enviroments. They have no memory of where they
> were, just reactive.

Responses to an environment are, in order, internal, local, distal,
tactical, social, representative, cognitive. These are inclusive
of each other. If you can't see why, wait until I'm finished my
next chapter.

> 5. Bipolar components is decidedly more difficult than using chips is
> untrue. -- furthermore, it is an opinion that isn't based on any
> empirical evidence whatsoever.

No, just more fiddly and time consuming, but at the advantage of
broader voltage ranges and direct motor drive.

> 6. There has also been espoused that devices like the Spider
> "learn to walk"

No, spider "anneals" it's processes quickly from random spacing
into a functional walking mode. "Learning" is too strong a word
for phase-space convergence even if it is adapting on the fly.

> Actual experiments with large circuits of nervous neurons
> show that the spacing between processes is remarkably
> constant.

No, only for cores less than 6. Experiments with process loops
larger than six show that a process can still be phase influenced
from the remaining charge of a neuron that passed 5 time cycles
previously. As the size of the loop goes up, the sensitivity of
the process is proportional to (1/(ln n/P)) - 0.4427), where P is
the number of processes, and n the number of neurons in the loop.
The effect is that the smaller the loop, or the larger the number
of processes, the faster the annealing time. What is surprising
though is that a 4 neuron microcore with a single process in it can
still be influenced by it's prior passing over a neuron 3 neurons
ago. It can "see" itself and be repulsed.

Try it. Dead short a cap without a process on it, remove quickly
and watch the process rate change in a microcore. Processes live
in a tenuous voltage space very susceptable to any charge
influence.

Neuron outputs trigger on the log decay of the coupling cap before
the inverter input, which is a very shallow voltage curve and
subject to any charge influences prior to it. Exact equations are
pending, but as many of you know, a 5% variation in coupling
capacitors can mean seconds in variation characteristics.

But you didn't hear it from me, cause you won't believe it until
you can use it.

markt.

----------------------------------------------------------------------
# sbolt@xs4all.nl # Steven Bolt # popular science monthly KIJK #
----------------------------------------------------------------------




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

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



Home