Freehand - an open source myolectric prosthetic hand

Thread Starter

strantor

Joined Oct 3, 2010
6,782
This thread is meant to be a companion to my blog, in which I will chronicle my adventure into designing an open source myolectric prosthetic hand. I will post a reply here each time I update the blog, to keep all the brilliant minds of AAC in the loop on my progress. This thread will also serve as an index to other threads related to the project.

Questions & comments are welcome.

Progress to date (1/3/14):
Very little.
Lots of ideas and a 3D printer.
and one blog post.
 

GetDeviceInfo

Joined Jun 7, 2009
2,192
hydraulics. Proportional valves abound. In similar fashion, arteries supply hydraulic pressure while veins are the return. Miniature proportional valves can deliver pressure/torque/speed in tight control. European systems tend toward high pressure, dense power systems.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
Someone sent me an email from the blog feedback box; I replied in the comments section on the blog. If that was you, check for your answer.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
hydraulics. Proportional valves abound. In similar fashion, arteries supply hydraulic pressure while veins are the return. Miniature proportional valves can deliver pressure/torque/speed in tight control. European systems tend toward high pressure, dense power systems.
Do you know of any suppliers of micro-hydraulic systems that I could use as inspiration? I need all of this to fit inside a forearm; an entire hydraulic system, pump, valves, electronics, reservoir, hoses, and not weigh 50lbs. Seems too tall an order to me, which is why I didn't consider it. But if it's out there, I'll look into it. I know Festo has some pretty small stuff.
 

GetDeviceInfo

Joined Jun 7, 2009
2,192
I've been looking, but haven't found anything yet. But, if you don't want it to be like all the others, you do need to chart new territory. Was it yourself that posted the link to Boston Dynamics. Many of their machines incorporate hydraulics as energy transmission.

Jet valves lend themselves very well to miniaturization. No reason that compensated pumps can't get down to literally any size. You need a prime mover regardless of the system you choose, so your weight is inherent. Not to mention the energy source you plan to draw upon. But the skeletal structure now becomes your battery, pump(s),reservoir, and actuator clusters. Cable cylinders become strand clusters. Wet hydraulics were used many years ago. You could supply with pressure lines, then allow the exhaust to flush the entire assemblies in a quasi hydrostatic system enclosed in a fluid bag.

But why mimic the limb, why not enhance it.
 

inwo

Joined Nov 7, 2013
2,419
In blog you talk about the choice between position mode and torque mode.

Is it necessary to pick one over the other?

It may be if this will be controlled by a simple direction command.

Seems the only difference between the two modes is feed back.

If you were to use the position feedback already in the servo for rough positioning, then switch to torque mode for grasping, it would be best of both worlds.

There would be a natural looking home position.

For grasping, a finger would first be moved to a computed destination, just passed the desired contact point, then as it met an "obstruction" it would continue in a torque mode until such torque as power to servo was reduced or dropped completely.

Torque feedback from sensors or current monitor.
 

#12

Joined Nov 30, 2010
18,224
Even if you use position control, the prime mover is going to have a power limit. I am just sitting back expecting you to integrate so many things in your mind and realize ways to do it because I believe you started with the right skill set.

1 vote of confidence.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
In blog you talk about the choice between position mode and torque mode.

Is it necessary to pick one over the other?

It may be if this will be controlled by a simple direction command.

Seems the only difference between the two modes is feed back.

If you were to use the position feedback already in the servo for rough positioning, then switch to torque mode for grasping, it would be best of both worlds.

There would be a natural looking home position.

For grasping, a finger would first be moved to a computed destination, just passed the desired contact point, then as it met an "obstruction" it would continue in a torque mode until such torque as power to servo was reduced or dropped completely.

Torque feedback from sensors or current monitor.
This all sounds a little more complicated than I think it needs to be. Addressing your comments I highlighted in bold, where is this position or "computed destination" coming from? These bio-signals do not come with a position encoded. They're just pulses, and groups of pulses, that vary with amplitude and frequency corresponding to muscle activity. They make less sense than what we digit heads would like. I think they will make every attempt to confound our best efforts at corralling them into neat position profiles so that we can make transitions from this mode to that mode, and return to such-and-such position afterward. That's why I want to forego the whole mess and and just simply amplify & rectify, and feed them straight to a current:force device.

It remains to be seen whether or not this can be done. I only have a cursory understanding of EMG signals and my plans may be way out to lunch. It may seem silly to forge head-long into something you have no idea will work, but ya gotta do something. This project is like a V8-powered merry-go-round with a stuck throttle. I stared at it from the bench for a good while before working up the brass to take it on. There's no good place to get on it; you just have to jump, somewhere, anywhere, and hang on. Something's bound to go wrong, and that's alright.

I am trying to tackle this beast from several directions at once and it just isn't possible. Right now I'm chipping away at 3D printing extremely durable parts and designing linear actuators. Mechanical design considerations and getting neck-deep into the world of EMG signals is still on the plate, yet to be touched.
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
Instead of hydraulics or pneumatics, have you checked into Flexinol wire (muscle wire) https://www.sparkfun.com/products/12096, Would make for a much lighter arm.
I do remember seeing that. It reminded me of the cape in the latest batman movie. I can't remember why I dismissed it. I did do some reading back then, but can't remember what I may have found negative about it. I'll do some reading again, and if I can't find anything negative, I'll give it a shot. That would be pretty awesome if it works as advertized "with great force" .

From the video, it doesn't seem to have a whole lot of force, but I think it should be considered as a single muscle fiber; and like a muscle is made of many fibers, there should be several of these in parallel. BUT, at 200mA each, it's going to get power hungry FAST, and as he mentions in the video, it takes a while to relax because it takes a while to cool down. And with several of these in parallel, I assume it will take even longer. And will get pretty toasty in there. It sounds like I might be retracing my previous steps that led to me discounting it as solution. Am I rambling? I'm sleepy. good night, thank you for the tips.
 

Metalmann

Joined Dec 8, 2012
703
"but I think it should be considered as a single muscle fiber; and like a muscle is made of many fibers, there should be several of these in parallel. BUT, at 200mA each, it's going to get power hungry FAST, and as he mentions in the video, it takes a while to relax because it takes a while to cool down."



From what I've read, I'd say you're right.

But then, I've always been an experimenter.;)
 

Georacer

Joined Nov 25, 2009
5,182
I really liked the testing procedure, and it's lovely when the results are predictable.

Did you calibrate your strain gauge beforehand? Did you know that it was linear, somehow?
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
I really liked the testing procedure, and it's lovely when the results are predictable.

Did you calibrate your strain gauge beforehand? Did you know that it was linear, somehow?
I did a two-point calibration in the meter, one at 0, and one at my body weight 210.5Lbs (95.5KG). The accuracy of the 1000Lb. load cell itself is so impressive that I count it negligible for my purposes. Ideally it should have been calibrated with 750-1000Lbs, but oh well, most of the measurements I'm taking are within my calibration range. The only other thing that could effect the accuracy is my calibration device: my bathroom scale that I weighed myself on. There's a rule in the calibration world; something like, a calibrated device can only be half as accurate as the device that calibrated it, regardless of accuracy spec. So if that's true, this laboratory grade setup is only half as accurate as my cheap chinese bathroom scale. and that's still good enough for me. I figure it's somewhere around +/- 10% (SWAG).
 

Thread Starter

strantor

Joined Oct 3, 2010
6,782
That's a very impressive load cell! How did it end up in your hands?
I don't know for sure actually. I used to buy used/unused obsolete stock lots of undustrial surplus, random stuff, motors, drives, solenoids, timers, etc. This could have come from one of those. Also could be from a company I used to work for that went through annual "make it shine" 5S/lean manufacturing stock purges. Those were happy times for me. Bins upon bins of good stuff, lined up to be hauled off to the dump, simply because it wasn't used often enough to meet the criterial required to keep it. I also accepted payment for work in spare parts a few times when I was self employed. I have so much crap I can't keep track of what I have, where it is, and where I got it. I didn't even know I had that load cell until I stumbled upon it looking for something else. Did not recognize it at all.
 
Top