Polymorph Hand
After Gareth's great walkthrough on making a servo operated polymorph finger I decided to make an entire hand. With the help of Gao, Claudia and my assistant Rose we made 4 fingers and a thumb.

Gao found that running a piece of brass rod through the finger made your holes line up and helped keep the finger straight.

I have since made the palm of the hand.

Since we want it to be able to pick up an LMR coffee mug I had to make the palm fold along two different axis's as you can see in the photo.This was done by running a soldering iron through the palm until it was 90% of the way through, then placing the palm in water and folding it along the line. The molten polymorph is squeezed out. When you open the palm again it is easy to dig out the excess polymorph.

The fold near the thumb was necessary for the thumb to oppose the fingers. Because of the fold at the base of the fingers I had to abandon Gareths mounting holes and just melt the fingers onto the palm. This was done using a gas torch (a cigarette lighter will do) to melt the surface of the palm and fingers where they needed to bond. Squeeze them together then hold in place while they cooled.

Looking at our goal I think I am going to have to build an entire arm....
With Gareth's X-ray photo for guidance I've decided to make a more human style wrist joint rather than the conventional robot style. This means I have to make polymorph ball joints.
Started with some polymorph balls and rod.

I then used the soldering iron to bore a hole in the balls. Because molten poly morph builds up around the holes I did this in two stages, placing the ball in cold water until it cooled and then using side cutters to cut off the unwanted polymorph between stages.


I then heated the tip of the rod with the gas torch before inserting it into the ball.

As the inside of the ball was also hot from the soldering iron you get a good bond effectivly making it into a single piece of polymorph.

I cooled the ball in water and cut most of the excess off with a sharp knife

A bit of delicate heating and smoothing with a wet finger gives me a fairly neat finish. I then cut the rod to length with side cutters and used the gas torch to melt the surface of the palm and the cut end of the rod before pressing them together.

Making ball sockets took some experimentation. After several attempts I found that the best way was to start off with a ball of warm polymorph about the same size as the ball joint.

Squash it slightly and then put it in cold water for a few seconds so it gets a skin that is cold enough not to stick to the ball. It will go slightly white. This is very important.

Press it gently around the ball. Work from the outer edge towards the center until almost 2/3rds of the ball is covered.

Once you have it evenly spread around the ball, run it under cold water until it hardens. Rotate the socket around the ball as it cools to ensure smooth movement. The end result should be a ball socket that fits firmly on the ball joint but moves smoothly and can be removed without too much force if required.

The thickness of the wall will determine how hard it is to remove or attach, as I do not want my hand to fall off when it picks up something heavy I found the wall needed to be 3-4mm thick.

Next I made wrist bones (ulna and radius). Solid polymorph would be to heavy so I rolled out a thin sheet and then tried to roll it up neatly. Easier said than done as it sticks to itself easily. My first attempt on the left didn't go very well. The second was better. These are reasonably stiff but still flex a bit.

I then heated the end of the bone and the outside of my ball socket and fused them together being carefull not to apply too much heat to the socket as I didn't want to damage the inside surface or warp it.



@ Thu, 2010-01-14 07:06
very punny
I gotta hand it to you OB and Gareth!, you sure know how to process digits.
I realise this post is not as humerus as first anticipated...
@ Tue, 2010-01-12 20:07
Great job!!!
Great job Oddbot!!!
I have one Question:where do you buy polymorph (claudiadagu@yahoo.com.cn)???
@ Wed, 2010-01-13 03:44
Yes!
@ Tue, 2010-01-12 12:49
Looks awesome
@ Tue, 2010-01-12 13:34
specs
@ Tue, 2010-01-12 00:53
Awesome work OddBot!!!! Will
Awesome work OddBot!!!! Will look forward to see it in action!!!
This is giving more cool ideas for articulations and puppets!! :-)
@ Mon, 2010-01-11 01:53
Instead of using servo
Instead of using servo motors to move your fingers and wrist, have you thought of using muscle wire. It mimics the movement of real muscles a lot more accurately than any servo would. You can connect multiple wires together for more strength and use an elastic (surgical tubing or simillar) as the tendon.
@ Mon, 2010-01-11 03:20
Muscle wire
@ Mon, 2010-01-11 06:54
If the fishing line is not consistent
You could try the braided fishing line. A lot more positive and predictable when it's under strain. If you need the springiness then you can build it in with springs like you mention. I just thought I'd mention that in case the nylon is inconsistent in it's stretchiness which I have found in a lot of cheap nylon lines. As in sometimes it stretches and never returns to it's original length.
@ Sun, 2010-01-10 23:52
Are you working ons omething
@ Mon, 2010-01-11 00:00
hehe
Heheh, your puzzling "off the wall" comment made me check you for hidden spam! But now I get it.
I second BoBBy's question: what are those lines on the whiteboard all about?
@ Mon, 2010-01-11 03:23
Ha ha, wondered how long it would take for someone to notice it
@ Sat, 2010-01-09 16:18
I wish I had the time
The fact is that this hand would need more joint and servos to do that and I don't have the time. I am just trying to make it pick up an LMR mug to demonstrate the uses and capabilities of polymorph (and for fun).
I would like to have one servo for each digit but will settle for 2 servos, 1 for fingers and one for the thumb. Not to mention the servos needed for wrist and elbow control.
@ Mon, 2010-01-11 00:31
Goldberg
This robot would be like something I would do. Most of my doodle concepts involve like 40 servos. Best of luck with it!
@ Sat, 2010-01-09 15:47
Number 5 alive!
@ Sat, 2010-01-09 07:01
Very impressive! Can't
Very impressive! Can't wait to see it grab something! :-)
It also looks like it would have great dexterity... like a hand for an amputee almost.
@ Thu, 2010-01-07 12:57
Would make a nice Halloween gadget
Nice work!
I was just thinking that it would make a nice Halloween gadget if you make it crawl on the floor :-)
@ Thu, 2010-01-07 06:28
Zombie Hand.
@ Wed, 2010-01-06 17:54
Agonist - Antagonist
Your Luck is in OddBot - my work is in the physiotherapy direction (would you believe)...............
Here are some basic ground rules to think about with regards to stable control of a joint....
Muscles mostly always work in opposing pairs.....
called Agonist & Antagonist.
This amazing arrangement allows for very fine strong control.
How to apply in Robotic sense......
The idea is to put 2 control lines on each finger or wrist joint
Then Drive both Servos so you pull the lines equally tight.
When you try to push the finger with your finger .... it will not move .... its under stable tension.
This is the base position.
To Bend the joint right .....
The Green Servo is the Drive servo, pulling and shortening the control line.....moving the finger to the right.
The Red Servo not only takes up the "Slack" but also places a resisting tension.....(this is the cool trick)
.....why do this you may ask.......
.... well if you dont.... (ie only use the green servo)
if you push the finger more to the right (with your finger) it is slack without tension..... and you can push it still further right.
If you do .... (ie use red and green servos)
if you push the finger to the right (with your finger) then it is not possible to move it more to the right because of the resisting tension of the red servo.
For completeness here is the opposite case.
Hope this helps - if the mechanics work then it will lead to a more stable control-able joint.
Maybe this would be a good idea for the wrist or thumb joints in particular...
@ Thu, 2010-01-07 05:13
Thanks for the info :D
I have already been thinking along these lines with one exception. I intend to use one servo because with a servo horn one side gives as the other takes. By attaching a strong spring on each cable to provide suitable tension the joint will always be held firmly in position. The only disadvantage of my design is that it can't "relax" and go limp. I will use this method for the wrist and bicep. The fingers for now will simply tense or relax as your prototype does.
@ Wed, 2010-01-06 10:56
Yes!
I want to try something different. So I will make an entire arm that moves in a similar manner to a human arm.
It wont be anatomically correct, after all the balls would normally be on the Ulna and Radius, not mounted on the palm as I have done.
@ Wed, 2010-01-06 18:18
Minor correction
MetaCarpels (base of hand) are plane or gliding
Thumb (opposable) is a saddle joint
@ Wed, 2010-01-06 10:47
Double wrist joint
Nice work!
Two ball hinges? Are you going to connect those to seperate arm bones? One ulna and one radius, so to speak? How elaborate are you going to recreate this skeleton?
(from wikipedia off course)
@ Tue, 2010-01-05 21:26
wow!
@ Wed, 2010-01-06 03:16
ASAP
@ Fri, 2010-01-08 15:12
Spinning Traces
You could also check these out - also from fishing shops.
They are thin stranded - normally stainless steel like strands- used for attaching "spinners" or even weights.
@ Sat, 2010-01-09 04:32
Good thinking
@ Fri, 2010-01-08 07:15
If you don't have any luck
If you don't have any luck finding heavier line, perhaps you could braid together lengths of the available light line? Not only would you get a nice durable 'tendon', but using bundles of smaller fibres is similar to what you see in real human tendons and muscles.
Awesome work so far, watching this one with great interest =)
@ Sat, 2010-01-09 14:32
Cross Tension
This is an interesting idea you bring up..........
If you bend one of your fingers - its automatic that the two neighbouring fingers move slightly too...
So if you had say 5 threads off one servo - 3 of these threads going to main control finger (ie middle finger) - then 1 thread going to index finger and one thread to the ring finger.
When you activate the servo - the main activation force goes to the middle finger and a minor force would go to ring and index finger. (more realistic)
However thinking about it........ this could be programmed in without the complex mechs........ so.... programming wins over mechanics.......
@ Tue, 2010-01-05 18:11
Opposable Thumbs
Nicely done, looks like youre steaming ahead - the Metacarpals taking shape too, nice solid base.
Isnt it amazing how you can increase and decrease the "flex" of polymorph (BTW nice tip re:- underwater trick)
Once the Bone structure is complete -
maybe some soft foam skin would be called for -
To help with the grip.
@ Wed, 2010-01-06 03:24
Definitely need to get a grip!
Nice photo Gareth!
Before I worry about the skin I am going to try and mimic a human wrist rather than the traditional method of one servo to rotate and one to bend up/down. Polymorph ball/cup joints will be needed I think. Want to do a workshop video for me showing how to make them :D
@ Wed, 2010-01-06 18:02
Hey you have found a suprisingly new technique
I thought that there would be no way that you could put hot poly onto cold poly (without them congealing together).....cool trick OB-1.
I was half way there (only in my brain though) - ie i was thinking about placing the ball part into the freezer for a bit....to make sure that it would stay really cold before appling the cup part, so there would be no secondary melting...........
@ Tue, 2010-01-05 18:02
That thing is scary!
@ Tue, 2010-01-05 13:36
eerie
@ Wed, 2010-01-06 03:14
servo actuation
Definitely going to use servos, we have some nice 13Kg/cm metal geared servos I want to use. Originally I was going to just use one servo to activate all fingers and the thumb but I am now thinking that a servo for each digit would allow for more possibilities.
e.g. hand signs like "thumbs up" or "peace"