Alt-BEAM Archive

Message #02257



To: JVernonM@aol.com
From: Bob Shannon bshannon@tiac.net
Date: Mon, 12 Apr 1999 19:46:57 -0700
Subject: [alt-beam] Re: Beam genome


JVernonM@aol.com wrote:
>
> Hello all,
> I guess the BEAM genome idea wasn't very popular. I got 5 responses from the
> list. One completely negative, two "interesting", and two corrections on the
> Z-bridge credit. So, out of 400 or so people I got two "kinda interestings".

I find the idea very interesting. I had not responded right away
however, having
been mentioned as one of the 'genome' developers (thanks by the way!).

If we are going to 'evolve' a better robot control technology, we need
to carefully
preserve the evolutionary process for study.

But we also need to be clear and what is new, and what is simply being
claimed as
being new. This is where misinformation and hype become major
obstacles.

> If this were one of Ian's polls, I'd say it went over like a lead balloon. I
> guess the idea occurred to me because the terms, genome, robo-genetic stock,
> and genetic algorithms are tossed around in Tilden's papers and this forum
> quite often. But, as far as my humble researching has discovered, there is no
> underlying structure to support those claims.

Heresy!

> I mean it's one thing to say
> you are expanding the evolutionary adaptation of the automaton, but without
> specific examples of how those changes are occurring and what characteristics
> are being used to promote and document that change, then you can't really
> speak in those terms. For instance, you can't point to a particular breed of
> dog and declare it a new more advanced breed without any means of comparison
> or agreed upon guidelines to support your claim of a new more advanced breed.

Exactly so! I totally agree. We do not even have a decent metric to
compare
the performace of the different solar engine circuits.

We have all probably read how the 1381 SE is 'more efficient' than the
FLED, yet
I've got a FLED Symet that can beat the pants off a ver similar 1381
based Symet.

Even the Beam Olympics I'm aware of so not publish any meaningful
performance
metrics. The recent photovore contest was great fun, and it did
directly bring
about new BEAM technology, but I think that a photovore race or obstacle
course
might start to show us which designs were 'better', rather than
'cooler'.

> You are required to follow guidelines that prove the blood line of that breed
> and how contributors have changed it into a better beast. I guess my point
> was, that if you have no official terms for comparison, and you don't
> incorporate a system by which you can show the evolutionary line that is
> leading you to that living machine, then the terms are meaningless.

Well, almost. We are not quite sure how Life began on Earth, but we at
least have
some criteria we use to recognise it when we find it. Because it has a
clear
definition, the term has a meaning.

What we need are some specific defintions and measurable quantities.

> They
> simply are fluff added to more fluff to make what your doing sound cool
> without any real documented approaches as are used to describe all other
> genetic trees. Without those procedures you're just saying the words. (Don't
> get me started on what robot parks have to do with the Jurassic period of
> prehistory, except for a cool blockbuster movie tie in.) I think that's the
> kernel at the heart of my musings. Tilden has presented his approaches as
> being high science, and a practical method for evolving autonomous living
> machines. But, there is no documented, scientific guidelines to promote and
> compare the evolutionary history of said machines.

I beleive that one key issue here is the scientific was to MEASURE these
things.

Without measurments, science itself is impossible.

> It also completely ignores those who have made some of those advances.

We already ~know~ that answer!

> I know I'll get flamed to death
> over this. That's what happens when you look God in the eye. I also realize
> that at this point our critters are usually elaborate toys, but I think many
> of us consider the possibility that the techniques may one day lead to very
> advanced living machines. If you believe that, I would think a more
> scientific approach would be the logical way to document that hypothesis.

I beleive that BEAM construction and design methods will advance
robotics, but
I cannot accept the idea that Tilden's BEAM approach will ever comppete
with CPU
based robotics. The difficulty is designing and evolving a hardwired
controller
can only be justifiable if the end result is sufficiently superior to a
CPU based
design.

This simply is not the case at all, a CPU can do everything a microcore
can do,
and then some. The first 'living machine', will have a CPU. Probalby
more of
them than the original Walkman has transistors.

But there is more behind the lack of science here. Without measurments,
we cannot
test the basic concept behind BEAM at all. Just how does it compare to
implementing
a stim/response based network with a CPU?

(No, a stimuli/response system running on a CPU is not "BEAM", just ask
a lawyer.)

As an engineer, I can tell you about the parts counts for each approach,
and the design costs. Where do these issues appear in this field?

> One more gripe here. I was leafing through the BEAM booklet the other day
> (yeah, I kept it even though it is a waste of money) and on page 49 there is
> a description of VBUG 1.5 WALKMAN next to a terrible Xerox looking photo of
> same. The description is as follows: "Single battery 0.7Kg. metal/plastic
> construction. Unibody frame. 5 tactile, 2 visual sensors. Control Core: 8
> transistor Nv. 4 tran. Nu, 22tran. motor. Total: 32 transistors." It says
> "motor" not 4 motors. I know the bot pictured has at least 4 motors. Is it
> any wonder Guinness got it wrong? I guess you could call it a typo, but I
> don't think so. And if it was why wasn't it corrected in subsequent
> publications (mine is the 4th issue)? Particularly when being documented by
> Guinness. You see, this kind of thing constantly crops up the more you look
> and it's very disheartening. It also delegitimizes Tilden and BEAM tech in
> the eyes of the robotic research community.

That's not what delegitimizes BEAM technology in my opinion.

Credibility comes from testable results.

If we want to evolve a living machine, we need to adopt a standard
platform.

I think something along the lines of Terry Newton's PIC BOT II is a good
starting
point. Rather than have many people investing many hours into learning
how to
free form the same six transistors, we should invest time into
developing new sets of behaviors for a standard platform.

Of course we would also have to have competitions, to test which sets of
behaviors were superior to other sets. Identical platforms with
different behavior sets could be placed into an arena (not a park) and
observed. Simple timed trials common in animal behavior experiments
could be used similar to the firefighting contests.

Has any BEAM design ever entered one of those? How did it do? How
would we know?

This is a much faster path to evolving a true living machine. It's the
cold equations
of modern technology. People stopped building hard wired controllers a
long time ago because of the economy of production and design. We use
CPU's now bacause they are superior in many respects to dedicated logic.

We can turn this same economy of production and design into an
evolutionary tool and develop a different form of living machine a lot
faster than current BEAM evolution.

And with measurable results.

> I'm not a scientist or a
> roboticist, but I know a scam when I see one.

Wellll, I wound not go that far. Some of this long haired AI stuff is
way out
there when compared to a bug's brain.

> By the way, when you do flame
> me and if you're a teenager, please be tactful. I'm a little weary of being
> scolded by 15 year olds. I have children older than that and I would dress
> them down severely for treating adults in that way. And I bet so would your
> parents.
> But fear not, God is still there, just don't look behind the curtain.
> Jim

Where is that little dog?

------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/list/alt-beam
Free Web-based e-mail groups by eGroups.com

Home