Alt-BEAM Archive
Message #11603
To: beam@sgiblab.sgi.com
From: Bruce Robinson Bruce_Robinson@telus.net
Date: Sat, 11 Mar 2000 08:07:00 -0800
Subject: [alt-beam] Re: smokeless h-bridge
Steven Dang wrote:
>
> What is the schematic of the smokeless h-bridge, or where can i find it?
Beam Heretics at http://www.serve.com/heretics/
Look in the Rigter Archive, and you'll see three variations on a 4
transistor smokeless bridge.
Bruce
11604 Sat, 11 Mar 2000 11:25:36 EST [alt-beam] Re: ARGGGGHH!!!! beam@sgiblab.sgi.com BUDSCOTT@aol.com I don't have a good picture of my circuit, but it is pretty much just the
almost complete walker scheme with cap values changed for longer timing. What
i did was add a photodiode and 1M resistor in a series parralel to the timing
resistor, this however has made the bot act kinda strange and only turn in
one direction, i've been flippin around the diodes and i cannot make this
behavior change! i can't really explain how it behaves too well, but for the
most part it only turns one direction.
Thanks in advance
-Spencer
11605 Sat, 11 Mar 2000 09:24:01 -0800 (PST) [alt-beam] Re: Any one form chesapeake or Va beach VA beam@sgiblab.sgi.com Max Inggal I don't axactly live in Va but live in NYC and know
how it feels to be one of the few BEAMers out on the
east coast and have friend that don't think much of
your bots. Sux I know.
__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com
11606 Sat, 11 Mar 2000 11:50:16 -0600 [alt-beam] Re: ARGGGGHH!!!! beam@sgiblab.sgi.com Brad Guillot The eeayest souution(?) would to go back to totally phoptphobic, and
reverse the leads of both the motors- this effectivly reverses the
photo-diodoes.
i think
brad
BUDSCOTT@aol.com wrote:
>
> I don't have a good picture of my circuit, but it is pretty much just the
> almost complete walker scheme with cap values changed for longer timing. What
> i did was add a photodiode and 1M resistor in a series parralel to the timing
> resistor, this however has made the bot act kinda strange and only turn in
> one direction, i've been flippin around the diodes and i cannot make this
> behavior change! i can't really explain how it behaves too well, but for the
> most part it only turns one direction.
> Thanks in advance
> -Spencer
__________________________________________
NetZero - Defenders of the Free World
Get your FREE Internet Access and Email at
http://www.netzero.net/download/index.html
11607 Sat, 11 Mar 2000 11:51:46 -0800 [alt-beam] BEAM CA "'beam@sgiblab.sgi.com'" Wilf Rigter Hi All,
Reading Jeff Williams email (Diodes in microcore) made me remember my own
introduction to BEAM, 17 months ago, when I was searching (dabbling) for
ways to build cellular automata (CA) in hardware form. I wondered if other
Beamers have had similar ideas and if Beam CA may acquire some life of it's
own.
The potential of using BEAM to achieve this goal was obvious: Simple neuron
like circuits that behave according to simple rules, interacting with other
neurons in networks to generate patterns of great complexity expressed in
the behaviour of real electronic and mechanical organisms.
So I plunged into the world (I should say jungle) of BEAM and immediately
lost sight of original goal. As with any learning curve, it requires some
time to become familiar with the language, history and tools of a new
technology but I quickly discovered that BEAM only defines (some would say
"redefines") a technical/artistic design approach or philosophy to building
robots. BEAM was originally a model or guide for robotics education and
competition. It really has no formal basis as a discipline and the
development of rules, methods and tools is soft/evolutionary. But it has
achieved a body of knowledge and an enthusiastic following that is the envy
of many other robotics groups.
The core of BEAM development and the current leading edge is based in the
BIOMORPHIC work of Mark Tilden at LANL with many contributions by others,
too numerous to mention by name. Currently, the majority of BEAM related
work is communicated through the mass media, via internet websites and
through the BEAM list mailserver.
Beam itself always was an experiment of a social, artistic, and technical
evolution of the Beam idea in which Beamers and their creations are the
cells or memes in a complex autonomous dynamic real world process with a few
rules, little structure, many degrees of freedom. Sort of like a giant
virtual Beam CA in a virtual Robot Jurassic Park neighborhood that Mark
peeks at every once in a while to see it some interesting patterns have
emerged, heheh!.
Anyway back to my beginning:
Cellular Automata, like Conway's Game of Life, are cells, in fact individual
bytes, living in a virtual community of a computer memory, governed by a CPU
executing rules that determine the life and death or continued existence of
each cell. This CA world can be observed through the looking glass of the
computer monitor as colonies of cells evolve into simple and sometimes
exquisitely complex groups one clock tick at a time. I found that small
colonies of CA, in small virtual worlds tend to settle into frozen, or
simple repetitive patterns but that some starting patterns are surprisingly
long lived. Moreover I found that adding structures to a CA world (CAJP!)
can provide templates that provide a far longer lived and more interesting
dynamic CA colonies even in a small colonies
The essence of CA is in the simplicity of the rules that govern behaviour of
individual cells with respect to their local neighbors and the emergence of
complexity in the behaviour of groups of cells
While it is obvious that CA with very large numbers of cells would have this
potential for complexity, I wonder if this would be true even when the
number of cells in a colony is relatively small. So I'm looking for an
algorithm which uses the minimum number of simplest rules to achieve the
maximum complexity in behaviour a minimum number of cells.
To put it somewhat differently, I am probably looking for rules similar to
those that define fractal patterns since this should yield infinitely
complex but bounded (interesting) behaviour. I was also thinking of a
3.14159265..(Pi) rule, an immortal CA (or bot) with behaviour determined by
each successive digit of the number Pi and which would therefore never
repeat. Interestingly, I believe a sequence of such behaviour would be
deterministic with respect to time but without reference to time zero, the
infinite never-repeating number sequence would yield behaviour as
unpredictable as chaos itself despite it's deterministic origin. The problem
that I see is that there is NO pattern in such a decimal number sequence and
it would not be interesting or aesthetic.
However, such a sequence which is infinite but which does not have patterns
may be an artifact of using decimal numbers for expressing transcendental
numbers. For example Minsky's simple iterative binary add and shift
algorithm as an approximation for sine and cos to calculate circles is food
for thought and would itself be interesting for determining simple robotic
behaviour
In the end,
my goal will be to develop simple Beam neural networks which generate these
iterative patterns using feedback connections based on simple rules to be
used to determine complex behaviour of robots especially when they interact
with each other and their environment.
wilf
> -----Original Message-----
> From: Jeffrey B. Williams [SMTP:jeffrey.williams@prodigy.net]
> Sent: Saturday, March 11, 2000 9:55 AM
> To: beam@sgiblab.sgi.com
> Subject: Re: Diodes in microcores
>
> Hello Everyone!!!!
>
> My name is Jeff and I've been lurking on this list for a while, I'm
> relatively new to BEAM, as I've only gotten into it since the Smithsonian
> article. First of all let me start off by saying I'm really impressed by
> the wealth of info available for a beginner like me on the web... as a
> result I've been playing around with a breadboard version of Chiu's two
> motor circuit... From what I've seen the PNC works fine... It only lets
> the core start up with one pulse...... Events later may cause another
> pulse
> to form, but at startup its pretty reliable. Honestly, I thought that
> this
> may have been a good thing... ( the extra pulse later on.) but I may be
> wrong if all it does is cause the bot to go frenetic with useless
> behavior... I've been messing with the microcore, and doing things like
> introducing pulses, and messing around with loads so that I can get a feel
> about what's going on. It seems to me that it is just as easy to get a
> two
> pulse net to collapse to one pulse as it is to get it to change to two.
> From reading Tildens and Brosls stuff ( I majored in Math and I am a
> geek ) , I believe that the core is always trying to get to it's lowest
> energy state, which is one pulse. It is only a slightly higher energy
> state
> in a four Nv circuit to have two pulses 180 dgrees out, and the core will
> stay that way if all is perfect. It is sort of trapped in an energy well
> that is above the level of the one pulse state, and that it will need a
> boost to a higher energy state before it can collapse to a one pulse
> state.
> My thinking is that , the clipping effect of the Schmidt triggers is
> enough
> to to take care of the differences in time constants due to the tolerances
> of the resistors and caps we use, and Chiu's circuit can stay happily in a
> two pulse state for quite a while... until motor loads start to
> ....(feedback, feedforward, implex????? Still don't know what to think
> about that one!) ..... will cause it to collapse back to one pulse by
> changing the phase seperation between them. It also appears to me that
> any
> sensor that has an effect on a Nv network will always have the effect ,
> directly or indirectly, of driving the network back to a one pulse state.
>
> In the end, I may be full of it, since I'm still on the breadboard at the
> moment. It looks to me that Wilf's circuit ( and man... is it a pleasure
> to find his stuff !!!) is really good at limiting the pulses to a one
> pulse
> state in the abscence of sensor input. A very good real world solution...
> as it keeps the core from having an epileptic siezure when a signifacant
> motor load causes the second pulse to form. It'll make the bot a much
> more
> determined walker and probably be a good start for a Sumo champion.
> I'm still planning on using Chiu's PNC as it is and add some sensors to
> the
> thing after I can get it fired up and walking right. Just in case though
> I
> have a momentary switch to hold one gate high so I can always take it back
> to one pulse.
>
>
> Freeforming, however is an art, and it is kicking my butt!!!!
>
> Peace,
>
> Jeff
>
> ----- Original Message -----
> From: Jim Taylor
> To:
> Sent: Thursday, March 09, 2000 12:54 AM
> Subject: Re: Diodes in microcores
>
>
> > You should know that Chui's schematics require some modification to the
> PNC,
> > and to the motor driver for correct operation. I'm not sure if he ever
> > revised his schematics. From my experience with PNC's, I'd say use
> wilf's
> > PNC free uCore.
> >
> >
> > > Thanks Bruce,
> > > I think I'm going to stick with the schematic Chiu has drawn and not
> try
> > to
> > > modify it to be PNC-less. I'll probably try that on another walker,
> but
> > not
> > > my first.
> > >
> > > So for the diode then, would a regular LED do the trick? Or do I need
> > > something special?
> > >
> > > Thanks again,
> > > Mike, AKA Flaming Headphones
> > >
> > >
> > > >Hi, Mike. That inverter in the lower left is the (now) old fashioned
> PNC
> > > >that extinguishes the extra pulse you get when the microcore starts
> up.
> > > >You need the diode to isolate the PNC after it times out.
> > > >
> > > >For a 4-Nv microcore, you can accomplish the same thing much easier
> with
> > > >a couple of diodes (and no PNC). Wilf came up with this a while back,
> > > >and you should find the circuit at the Beam Heretics site.
> > > >
> > > >Bruce
> > >
> > > ICQ: 17940917
> > > E-mail: headphones99@iname.com
> >
> >
> > |___|
> > -------O()O-------
> > James Taylor
> > URL: http://fly.to/springmeadows
> >
11608 Sat, 11 Mar 2000 12:39:26 -0800 [alt-beam] The latest websites (and Wilf's circuits) alt-beam@eGroups.com "J Fisher" I've been out of the loop for a while, and I'm having difficulty
finding the bounty of the last few months, due to my heavy reliance on
sites like Fang's and Beam Online, which are currently in a slow period
regarding updates. Are there other sites that now fill the same
function?
I'm particularly looking to see what kind of circuits have been
developed over the last few months(/year...). I know Wilf's done some
neat stuff, and I assume
that other nifty, odd, esoteric, specialised, complex, simple, and
interesting circuits have also appeared. Is there a site or group of
sites that collect this stuff? (If it doesn't exist, a giant circuit
site that's got all the stuff in addition to the ubiquitous 1381 SE,
FLED SE, photopopper, and other stuff, would be a great idea, hint hint
:-) I've got a couple of circuits that could help get it started... :)
Home