email@example.com (Bruce) wrote in message
Does not have to be servo pwm. I am running coreless motors at 30 KHZ.
look up the motorola 56807. The 12 PWM channels are quite powerful. 16
channels of 12 bit a/d. It's definately not a toy, unless you want it
I'm using 6 of their MiniPod's to drive the 12 motors of my centipede,
with canbus to connect. could have used 1 IsoPod, but the wiring
demanded distributed processing.
I don't know if your architecture demands PWM control or analog (
which you can get from PWM), or if you need fast floating point stuff,
multi-tasking, etc, but this product has a lot of potential.
I have a project I am using it for at work, that reads 6 quadrature
encoders, using encoder pulsewidth for evtremely fine velocity
feedback, canbus, A/D, both serial channels. Lots of floating point
There have been many requests for links/pictures/sources/websites from you
on this thread. Can you provide any? Or is this vaporware? Sorry, not
trying to be agressive. It's just that you are making some pretty wild
claims. How about something to back them up?
Here's a short list of what you said:
"The SynthMuscle I already have and can produce 30kg pull at
speeds up to 40cm/second per "chord"."
"The arm is already show usable weights up to 15kg."
"Every SynthMuscle has an integrated 20MHz microcontroller and
"For each "joint", there is a larger 20MHz
"For each "limb" (torso and neck also considered limbs),
there is yet again another 20MHz microcontroller"
"Linking all of these microcontrollers is a branchable AFAP bus."
"For higher functions, there are FSys boards. These are 2GHz AMD
"In all, there are 230 microcontrollers and 21 FSys boards"
"The pure electrical version is really so simple "..."But, it works very
"The electrochemical version quickly degrades. But, once upon a time I
didn't ever think I would get the electrical version working."
"the current hurdle (ion migration...self contaminating)"
"total system through put is 134Gbit/sec), the FSys C compiler, buses,
etchmasks, power systems, and the whole kit and kaboodle."
How about it? It's OK to dream. That's where things come from. You are
implying that you have things running. The project you are proposing is
monumental. Tough to do for a well-tuned team, probably safe to say
impossible for one guy to get beyond embryonic stages. What do you REALLY
have, if anything? A few pictures, links and MPEG videos will do fine.
Eagerly awaiting to see what you got, and, perhaps, participate, if it's
Whereas I tend to agree with your disbelief about achievements
and claims, I disagree with you about the one/many requirement.
One born natural does work of this nature that otherwise requires
teams of trained mediocrities. It is difficult to see why, if the guy is
so much, that he needs to take on a team to do the rest. He is apparently
claiming a motive improvement that alone would make him his fortune
if he patented it.
How often have you seen a technical leader move on and have to
be replaced by a team?
This illustrates a lot of what is wrong in the AI and robotics world - too
many jumping on the bandwagon and basking in the reflected glory
Sorry, I tend to assume people understand just how complex a real humanoid
Let's see, every major university in the world has been working on various
aspects of the problem for decades. Nobody has gotten a single bipedal
walker to walk reliably, much less run, jump, etc. That right there is
enough to keep a group of very smart folk busy for a while. You think that
Asimo walks like it just got an enema 'cause they think it's funny? Nope,
they can't figure out how to build a machine that will walk dynamically,
like we do.
Let's not even mention the chemestry/physics genious required to figure out
how to power this thing. If I extrapolate from what I've read so far, each
one of these "SynthMuscles" requires up to 15A. It sounds like, if this
thing is walking around, the EMP pulse alone will render you sterile. :-)
I'm sorry. I'm poking a little fun at this whole thing. Maybe I shouldn't.
I don't know, to me it sounds like someone who's thinking out loud and
hasn't built anything at all. It sounds like a little bit of science with a
lot of science-fiction. In the real world a project like this would
probably cost $10 to $50 million and take at least a dozen people to
undertake. And many years to completion. Maybe decades.
Ya know, a few links, pictures and videos could support the claims very
nicely at this point ... I'm surprise the conversation continues and nobody
is holding this guy to the claims made. At least a couple of people asked
for links, pictures, documentation and they were sumarily ignored.
So, what you're saying is that there are large numbers of
teams and peoples who haven't got a clue. When the one man
who solves the issues appears, that one man is the important one, all
others are bandwagon jumpers.
Martin is right, with a few exceptions. Everyone is trying to do
the same thing, but separately (which is only as great as any one
group). Universities are only after what the particular students
thesis is based on. Companies are trying to make it profitable and
only go as far as the immediate off-the-shelf parts can be fit
As for the "one man", I think it will be a dozen men that step
forward with an answer and their combined answers will appear to be
one man. Think of history...Edison had dozens of talented men behind
him who solved the problems. Edison just asked a good question, but
didn't have the solution until after over 1200 failed attempts.
You are right. Airy seems to have a really low opinion of team-oriented
workflows. I don't. I used to, to some extent, but, all you have to
do --once-- is try to tackle a project that's bigger than you (no matter how
good you are) to learn that lesson. Besides, who wants to be a loner?
As a point, it can max at 45amps per muscle...six coils in three sets
- 15amps max...sizzle, sizzle. Actually, it should never reach the max
otherwise it needs to sit in the smoking section. :o)
Also, is anyone good with nailing down leakage of a magnetic field?
Martin is right that the thing broadcasts like microwave. Multiply
that by all the SynthMuscles we will be needing and sterile might not
be far off the mark.
If you think that is all there was to it, then why would I bother
asking for help??? (my wife wouldn't quite go for the handmaindens,
There are so many problems to overcome.
One man means nothing! It takes a team.
Everyone is probing and nervous. Nobody wants to give away
anything and as such, it goes nowhere. So here we are.
The Team would really like to know more about your brain modeling
(I assume you are using some version of a perceptron). We would love
to learn about you mechanics. We would love exchange ideas on
We could really use your help. I have spent years working on the
vision and frameworks. I have nothing other than a blank spot with an
FSys board assigned to connect the two.
If George 1.0 can give you the world from a steroscopic view (even
if everything was anonymous objects) and give you the mechanics to
move with force (offloading much of the processing to the
microcontrollers), and give you the audio and where is coming from,
what could you do?
Seriously, think about it. What could you do? Or more
importantly, what can you do by yourself compared to what all the
talent that is silently sitting by could do with you?
The only stupid idea is that we are islands that can take on the
EXACTLY!!!! That is the problem. It is going NOWHERE!!!
Everyone is so afraid of not being perfect, they won't speak up.
There are so many talented, imaginative, and brilliant people just
standing around waiting.
Well, I don't have all the answers and I am so very, very flawed.
I am even dumb enough to think this might work. I'll even give you one
better: I was born and reered in a small coal mining town deep in the
Appalachians...an official Hillbilly. Who am I, nobody, just me.
But, I am going to try. Hopefully I can contribute in the areas I
have worked so long and learn more from others.
I know the nature of people and the ridicule I will at some point
receive, but I am standing up and trying anyway. You think there won't
be problems...I think at several time it will appear hopeless. But, I
What about you? Do you want to help or stand on the sidelines safe
in veil of "can't happen superiority"?
Oh Ye of little faith :o)
I'd be signing up for this if I thought he needed a nuts, bolts 'n' levers
man, and I wasn't already doing one of my own.
OTOH I still think it warrants a cool domain name and web site.
Robin G Hewitt
There is plenty of room for all kinds. And it is suprising that
not only do we need the "nuts...and bolts :o)" types, sometimes
projects delay over such things as naming conventions.
Everyone is most welcome. You mentioned domain naming. Do you have
a better name than George 1.0 (I am not a marketer...I have no
imagination in that area)? Some team member have expressed that it
"sucks" (and I agree totally). Join in an help name the thing if
If you only new how anxious I have been waiting for someone like
you (the fishing phase...I hope :o). I'll let you, and everyone else
decide if it is vaporware. Like I stated, I am opening myself up (and
will be taking black eyes I'm sure ;o).
There is a group on Yahoo called George 1.0. I have been posting
stuff there as I get it cleaned up (from scattered notes...it is
trickling in) and as people are weeded out. The group is inclusive of
everyone, but there are those that do not want to be as deeply
involved, but still can contribute. I will be posting pics there this
weekend. The only requirement to join is you need to tell the Team
about your experience, skills, and area you are interested in.
You speak of a team...that is what I am after. I have spent a
great deal of time and money getting things to an enabling point. My
goals are to walk a group of people through the initial design of
George 1.0 giving them the pieces and letting them point out problems
and contributing to the design and assembly. You are right that it is
too big for one person.
So where am I at: embryonic and going forward.
1) I have a working electrical version of a SynthMuscle being
integrated into a robotic arm for a friend who is quadraplegic. While
integrating it, I realized that I had a usable SynthMuscle and decided
now was a good time to start bringing the right people together. This
is not rocket science...the SynthMuscle is really, really simple. It
needs a better strain gauge.
2) The next batch of boards to be etched will contain some to be
sent to team members. Also, SynthMuscle chords will be sent as well.
Who gets them first is yet to be determined. They'll be the judges.
3) I have dozens of failed SynthMuscles...pick your flavor. It
isn't like I made two types, one works, the other doesn't. I have
failed many, many, many times. ;o) But this one shows promise.
4) For the working SynthMuscle, there are two types of control
boards: the first is power MOSFET based and the second uses IGBTs. The
microcontroller is a PIC16F628. The control is simple: pull this
hard/fast or let out this hard/fast.
5) For the Joint controller, it is a PIC16F877 with the support
for the AFAP Client bus. The Joint controllers job is to monitor the
strain on the SynthMuscles and control position. They receive an
equation of motion (with ranges of course) and a time/rate frame to
6) For the Limb controller, it is basically the same board as a
Joint controller without using the ADC. The current limb controller
uses a National 9603 USB chip to interface to a PC as a PID. The
reason for this is he already has a mouse (specialized, of course).
Mounted to his chair is a notebook computer which acts as the gateway
between the two devices so he can make changes himself (VB
program...mouth stick to program).
7) The test FSys board has been etched. It uses an AMD Athlon
processor without support hardware (no north or south bridge...yes,
there are a lot of pins :o). The board uses an eeprom to boot, copying
the program to the local SRAM. There is a single port which consists
of two Cypress dual-port SRAM sets used as the interface (one
read/write, the other read only). I gave maximum throughput, which of
course, isn't the real speed. But the bus is 64bit and I'm sure you
can work out the math.
8) The FSys C compiler is together...loosely. It has just enough
coding to compile ANSI-ish C, assemble the code with enough bootstrap
to get the program in the SRAM and test the porting.
9) For the eyes, I use OVT7620 cmos imagers. If you are familiar
with them, you will know what a great chip it is. The board has the
imager mounted on front with a control pic and support mounted to the
back. The OVT7620 is programmed through an I2C port which is native to
a PIC. The output is for SRAM, but also has a debuggable NTSC output
10) The interconnect hardware for the eyes to the FSys board is
in design only (spec'd, layout finished, not etched). Basically, the
support just loads the data stream from the eyes.
11) For the 230 microcontrollers and 21 FSys boards, the design
of George 1.0 I have been putting together requires that amount. I do
not have all 230 microcontrollers or 21 FSYs boards yet. George 1.0
isn't built yet. If he were, why would I need a team to help?
12) For the "bones" of George 1.0, plastifoam is shaped with a
hot wire, then wrapped with 10.9 oz carbon fiber and epoxied. The
joints are made from machined UHMWPE. UHMWPE is, of course, not
chemically compatible, so the connection is a mechanical one. This is
what I have used in the arm and it works, but there are limits.
13) The vision 2D-3D integration is not finished. The 2D vision
system has been created over the past 7-8 years and does not work like
anything else you have probably seen. This is the one area I will have
the greatest difficulty in disclosing because it is in production and
a lot of people count on it for their living. There are also a bunch
lawyers involved and contractual obligations. We'll work something
14) The audio system has had no work done to it. There is a guy
on the team who asked for this system above all else.
15) The balance is going to use three 3-axis accelerometers and a
2-axis polymer inclinometer. Currently, I have a test board checking
out the ACH-04-08-05 chip using it conjunction with an old 2-axis
inclinometer (I don't even have a part number for it).
Ok, I probably missed a dozen things, so fire away. Also, if you
are a team player, check out the group. Remember, this is a start: you
are not going to get a complete spec book of George 1.0. As we cover
each topic and I have time to clean up my crappy notes, things are
going to adapt.
OK, I'll bite. I just submitted a request to join. Let's see what you got.
I'm getting done with a huge project (two man-years). If it shows promise I
might consider getting involved. I have your 200+ microprocessors sitting
on a shelf. Including many hundreds of MOSFET's and related components.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.