Advanced robotics

Hello everyone,
Is anyone interested in working on actual utility class android (~5ft high)? I am looking for people with skills, but newbies are also
welcomed (as long as you can function in a team). Also, forget the gears and electric motors, this is SynthMuscle domain only (have a SynthMuscle working and currently integrating it into a robotic arm for a friend who is quadriplelgic...25 integrated processors).
No B.S., no wishful thinking. This is a very big project and it needs to be divided to conquered. Think of this as the "Shareware of Robotics"...or simply as "George 1.0".
Several subassemblies are in progress: 3d steroscopic vision processing, triple-3-axis accelerometers for balance, audio spectral analyzer for voice recog., synth muscles w/node controllers, central bus (you might not think this would be important :o), etc.
The architechture is simple and adaptable and subject to change.
Decide an area of interest and let me know. Sincerely, Bruce
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Sounds interesting... can you reveal where you get these SynthMuscles?
- Daniel
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Daniel, I make them. They are not overly complicated, but the math that enables them to work required some very specialized materials. Also there were some power requirements that only recently enable it to function.
As for revealing where...how about one better. As I put the stuff together (converting my scribbles into a legible form ;o), those part of the George 1.0 team have access to the source, designs, schematics, etch masks, and some components are going to be shipped between members for experience and refinement.
Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Bruce,
Sounds very interesting, but what are your projected power requirements?
--
Shawn


"Bruce" < snipped-for-privacy@yahoo.com> wrote in message
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Hi Bruce
I just found my original post and your answer in someone elses reply.
My android is 5' 4" and 35kg. The height being optimised to fit the motors. To get the necessary ft.lbs of torque I use good quality motors and low loss gearing. It is far from slow and feeble :o)

Processing power is wonderful but it doesn't pump much heat :o)

I've been looking for a small sized mother board with credible power requirements. Have you found something nice? The pack of 386ex processors in permanent residence on top of the kitchen cupboard is starting to look attractive.
best regards
Robin G Hewitt
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Robin, If your droid is 5'4 and only weighs 35kg using gears and electric motors and not being slow and feeble, then my hats off to you. There are certain limits mathematics places on designs (at least, I am bound by those laws of physics ;o). I once did the math using PacSci PowerMax II and API Motion steppers/brushless motors, and the numbers were not very pretty. How exactly did you get around the math of it???
Heat disipation takes place over the entire surface of the droid. The materials allow a very high operating temperature. The thermodynamics (environment vs. source) function within spec. To supplement cooling, there are low-pressure air lines that run to all the SynthMuscles. Cooling is like a man walking a high wire...;o)
The only location that has a real heating problems is with the FSys boards. They are liquid cooled through small flat pack mounted directly to the processors which is routed through a small "radiator" placed at the air intake in the chest.
Unfortunately, I have to make the FSys boards. I also wrote the C compiler for it (when time permits, making a fully C++ compiler). The boards are updated by using a SmartMedia card (simple interface) that is placed into the "head". There is a PIC that reads the card and flashes the different eeproms by file name.
Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I thought 35kg was heavy ! My approach was much less scienterrific, I weighed a few components, added them together, then added a bit more for luck. I reckoned it was essential to keep the weight down if there was to be any hope of alacrity. I have avoided stock components, treating each joint as a separate puzzle that needed solving, that way there is very little wasted metal.
OTOH, perhaps feeble is a relative term, I must admit my design is more life sim than killer cyborg ;o)
best regards
Robin G Hewitt
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Shawn, Considering that each SynthMuscle cord has a max pull of 45amps (ouch), the power spikes are none the less havac forming. The microcontrollers used in the SynthMucle uses a 40kHz PWM to limit current. Basically, the system uses the absolute minimum power to sustain a position. This is done by the interaction between the joint controller and the synthmuscles.
At first, to solve the spike problems seemed uncrossable. Luckily the military came to my rescue. Carbon aerogel capacitors met my requirements of weight, size, and capacity. This may not seem important, but in motion there is a MAJOR difference in steady power and initial movement power.
As for a total system power, put simply, it is >>>HUGE<<<. The SynthMuscle only has about a 40% efficiency (but not nearly as bad a most poorly operated steppers or electrics). There is actually a two stage power system: the first is the stored/filtered power from lithium ion batteries (with a custom controller), and the second is a small magnetic-bearing based microturban (yes, hacked straight from a old backup power supply...old meaning someone damaged the case and other components but the generator and turban was intact).
Again, this is a utility class android.
Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

It will certainly be a comfort during the Winter months. Obviously I am misunderstanding SynthMuscle. I thought you meant SMA smart wire, what the heck is SynthMuscle?
best regards
Robin G Hewitt
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Robin, Yes, you've got it exactly. The pure electrical version is really so simple (once you have the right materials) that it is kind of sad. But, it works very well (thanks to the math and scale...long story). It does require constant control from a local microcontroller which is the reason for the special requirements and VERY large number of microcontrollers.
The electrochemical version quickly degrades. But, once upon a time I didn't ever think I would get the electrical version working. In the design of George 1.0, the SynthMuscle can be swapped out with very little modification to the rest of the system if/when the electrochemical version gets past the current hurdle (ion migration...self contaminating).
Until then, George 1.0 is being put together purely electrical. I have been looking for others to join a team to fill in the vast gaps (I am focused on the utlity and base function, which makes for a very boring droid). My job is to provide the underpinnings: building the basic frame and SynthMuscles, making it walk and manipulate, giving a real-time 3D object recognition, and the power system to make it usable. The Team will take it to where it needs to be, which is where I can't alone.
This is a Team effort.
Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

So, presumably you've patented it, no? You'd be nuts not to.
-----sharks
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Sharkey, I have been through the patent process many times (some completed, some pending), but this one I am going to be giving away. And yes, maybe I am nuts.
And if that isn't enough I am giving away the diagrams, specs, schematics, source code, higher level processing functions, ultra high power board info (each with a 25Gbit/sec data processing throughput...total processing is 537Gbit/sec), the zero delay interchanges for the FSys boards (6.4Gbit/sec data exchange per port...total system through put is 134Gbit/sec), the FSys C compiler, buses, etchmasks, power systems, and the whole kit and kaboodle.
So, is that more than nuts enough? ;o)
The whole point is to take this where no one is willing to go. Everyone is so guarded and secretive, afraid they might give away the grand secret and loose out on a zillion dollars. I say B&*@#$&T! It takes more to make money than a good idea. It takes an industry.
That is the whole point of George 1.0. It is open source, highly advanced robotics. It can walk, run, jump, dance, and most importantly

Bruce
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Bruce wrote:

As long as this is open source (and there is a licence to back this up), I'm willing to help. -- D. Jay Newman Programmer, Writer, Gadgeteer http://enerd.ws/robots /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
ISTR George 1.0 could also make the tea when it ran the ICL1900 systems. Much, much easier than scheduling jobs yourself!

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Airy, First there is a ton of work, then the tea comes later ;o)
Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Jay, I have been "hacking" a good general open source license. It gets a little hairy when hardware is added (especially things like the SynthMuscle). When it gets to a good point, I'll get the lawyer to through it (he'll freak when he sees what I am about to give away :o).
I take it you are interested?
Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Bruce wrote:

So far it sounds interesting. I'm doing work in behavioral programming and sensory fusion, so it sounds like a good match.
Do you have a web site? -- D. Jay Newman http://enerd.ws/robots /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Mike, Thanks for info, but this doesn't use servos. The reason for so many PICs is due to the requirements of the SynthMuscle and the constant readjusment needed. Also, there is a very complex interaction between the SynthMuscles and the Joint controllers that requires a ton of processing.
What else you got? I'll take anything that can make this thing simpler as long as it meets spec. ;o)
Thanks, Bruce

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Thu, 28 Aug 2003 22:30:19 -0400, Bruce wrote:

You might want to take a look at Atmel's AVR's - there's bound to be a device in their wide line-up that will meet your needs, prices range from $2 for lower end up to around $10 for higher end parts, depending on where you buy. Typical parts run at 16 MHz and are true RISC, i.e., 16 MHz = 16 MIPS. Another big plus for you is that the AVRs are supported by the GNU C/C++ compiler (GCC), the core of the defacto standard open-source tool chain. The AVR port (AVR-GCC) is quite nice and a very complete standard C library is also available (the AVR-LIBC project). Also, it seems like everybody and their brother are using AVRs these days so there is plenty of knowledge and help available from the on-line community. See http://www.avrfreaks.net/ for starters.
Many folks offer PCBs and/or completed boards - I have a couple available here based on the ATmega128 (top of the line AVR):
    http://www.bdmicro.com/
I've got another board coming out soon which will be smaller (2.0 x 3.6"), and include RS485 networking on-board. Protos aren't back yet, so I don't yet have a firm date for availability.
You most likely wouldn't need the ATmega128 for all your processors, though. One of the smaller ones would probably be plenty, depending on how many I/O pins you actually need per location/joint.
For the full parametric data table listing Atmel's full AVR line-up, see:
http://www.atmel.com/dyn/products/param_table.asp?family_id `7&OrderBy=part_no&Direction=ASC
Best of luck!
Cheers, -Brian
--
Brian Dean, snipped-for-privacy@bdmicro.com
BDMICRO - Maker of the MAVRIC ATmega128 Dev Board
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Fri, 29 Aug 2003 09:28:21 -0400, Bruce wrote:

Not sure why it needs to be 20 MHz, but you should take a close look at the ATmega8 (16 MHz) - it's got 3 hardware PWM channels which should easily meet your output requirements and not load down the processor at all, leaving it free for other processing. I'm not sure what "AFAP" is. It's available in a 32 pin TQFP package so it can mounted on quite a small board and you would have plenty of I/O available as well as A/D converters, I2C, and lots of other peripheral capabilities. They only cost $3.75 each from Digikey (qty 100). A freely available (and very nice) C Compiler is available with Gnu's GCC which can be hosted on every major operating system like Windows, Linux, FreeBSD, NetBSD, etc, etc:
    http://www.gnu.org/software/gcc/gcc.html
Thus, your implementation would truly be open like you are wanting to make your design. Sounds to me like it fits nicely with your project goals. What could provide a better fit?
Cheers, -Brian
--
Brian Dean, snipped-for-privacy@bdmicro.com
BDMICRO - Maker of the MAVRIC ATmega128 Dev Board
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

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.