Ring shaped potentiometers

[Alptraum]

Hi,
I have wanted for a while now to build a simple robotic arm which follows human arm motion, via a control interface which would go around the arm and move with it. Although the shoulder and elbow movements can be done using normal pots, the most efficient solution for wrist and upper arm twist would be a ring shaped potentiometer which would go around the arm - the upper half would be fastened to the shoulder  and the lower half fastened to the elbow joints (which are always perpendicular to the elbow bend direction).
The only problem is, no matter how much I search, my google-fu fails me on this one.
Anyone got any ideas? Is it possible to make one of my own?

Any help is much appreciated.

[paulstreats]

maybe something like this? http://www.sparkfun.com/commerce/product_info.php?products_id=9074

[Alptraum]

maybe something like this? http://www.sparkfun.com/commerce/product_info.php?products_id=9074

Kind of, but I need it to be bigger - it needs to fit around an arm. If it already has a metal casing as well, instead of being a board, that makes my life a lot easier.

[Alptraum]

Actually, the linear versions that they sell could easily be made into what I'm after (provided they're flexible). Thanks for your help!

[Metal Slug 2]

Your current project is the next one on my list!

Rather than worrying about rotary pots, I am going to use flex sensors from Sparkfun:  Link

If you can afford them, they work great.  Although be very careful with them, as they are rather fragile at their connection points (which btw are breadboard friendly).

Follow step 4 of THIS Instructable on how to setup (and protect) the flex sensors.  If your not comfortable with soldering directly on the connections of the sensor (like me), I found that they fit nicely tightly in between these Ic sockets.

I would suggest the use of these flex sensors at the elbow, one or two at the wrist, and perhaps one for a claw (just one sensor attached to thumb or middle finger).  Or if your rich, or just insanely determined on building a kickass robotic arm, maybe even a flex sensor for control over each individual finger...

Now the shoulder joint movement along the pitch axis would be a tricky thing to measure with a flex or rotary potentiometer, so I'm planning on mounting an accelerometer (or gyro, if i can find a cheap Wii Motion Plus somewhere) on my lower tricep, near my elbow, and just measuring the angle.

[Ro-Bot-X]

[youtube]qLqq5EulHBk[/youtube]

[Alptraum]

[youtube]qLqq5EulHBk[/youtube]

The awesomeness of this knows no bounds.

Your current project is the next one on my list!

Rather than worrying about rotary pots, I am going to use flex sensors from Sparkfun:  Link

If you can afford them, they work great.  Although be very careful with them, as they are rather fragile at their connection points (which btw are breadboard friendly).

Follow step 4 of THIS Instructable on how to setup (and protect) the flex sensors.  If your not comfortable with soldering directly on the connections of the sensor (like me), I found that they fit nicely tightly in between these Ic sockets.

I would suggest the use of these flex sensors at the elbow, one or two at the wrist, and perhaps one for a claw (just one sensor attached to thumb or middle finger).  Or if your rich, or just insanely determined on building a kickass robotic arm, maybe even a flex sensor for control over each individual finger...

Now the shoulder joint movement along the pitch axis would be a tricky thing to measure with a flex or rotary potentiometer, so I'm planning on mounting an accelerometer (or gyro, if i can find a cheap Wii Motion Plus somewhere) on my lower tricep, near my elbow, and just measuring the angle.

I don't fully understand what you mean by the pitch axis at the shoulder joint... I mean measuring the amount by which the lower arm twists in relation to the upper arm, or how much the wrist twists in relation to the forearm.

Also, since your entire arm might stay stationary in relation to your body but move in relation to the ground, an accelerometer will give inaccurate readings - unless you give it another reference point, or have another accelerometer mounted elsewhere on the arm which would tell the microcontroller what the 'neutral' position is. The other potentiometers would then be measured in comparison to that, and that would (hopefully) negate the effect.

Does anyone know if those linear potentiometers mentioned in the 2nd post are flexible? If I can get them to fit around my arm, then I can build a rig which keeps the strip stationary but moves the stylus along it as I turn my arm..

[Soeren]

Hi,

Also, since your entire arm might stay stationary in relation to your body but move in relation to the ground, an accelerometer will give inaccurate readings - unless you give it another reference point, or have another accelerometer mounted elsewhere on the arm which would tell the microcontroller what the 'neutral' position is.

Didn't you ever try a Wii?

[Alptraum]

Yes, a Wiimote could work - but only if the only thing that moves is the arm.

e.g. -
Say you hold the Wiimote at arms length, standing straight upright, with your arm at 90 degrees to your body.
The remote will read as if it is being held horizontally.
However, if you then bend your upper body at the hip by 45 degrees, the remote will no longer be held at 45 degrees. Thus, even though your arm is still at 90 degrees to your body, your upper body (and by extension, your arm) is then at 45 degrees to the ground.
Thus, the computer perceives the remote to be at 45 degrees.

The movement will thus not be accurate, as the reference point for the sensor is the ground, not the upper arm. Thus, it will not measure angle to the upper arm, but to the ground.

Unless I've got the wrong end of the stick...

[Metal Slug 2]

Your absolutely correct.  And if you want to build a robotic arm that mimics arm position in relation to the body via two accelerometers, that would be perfectly fine.  Its really all personal preference. 

Theres pros and cons for both methods -1. robot arm position relative to the ground or 2. robot arm position relative to your body.

Why not try incorporating BOTH methods, and being able to switch between the two?
Better yet, try adding a third alternative where you use a potentiometer to mimic the angle that would otherwise be measured by the accelerometer mounted on the body.  That way you could possibly be lying on your back, while your microcontroller will think you are upright.

[Alptraum]

Your absolutely correct.  And if you want to build a robotic arm that mimics arm position in relation to the body via two accelerometers, that would be perfectly fine.  Its really all personal preference. 

Theres pros and cons for both methods -1. robot arm position relative to the ground or 2. robot arm position relative to your body.

Why not try incorporating BOTH methods, and being able to switch between the two?
Better yet, try adding a third alternative where you use a potentiometer to mimic the angle that would otherwise be measured by the accelerometer mounted on the body.  That way you could possibly be lying on your back, while your microcontroller will think you are upright.

So you mean that the user sets the neutral angle manually, or is the pot connected to some part of the body?

[Soeren]

Hi,

e.g. -
Say you hold the Wiimote at arms length, [snip explanation]

The movement will thus not be accurate, as the reference point for the sensor is the ground, not the upper arm. Thus, it will not measure angle to the upper arm, but to the ground.

Ah, now I see what you mean.
Well, a Wii remote (without the plus addition though) and a nunchuck can be had for like $15 from some Chinese sites, so it's cheap to experiment (I'm soon ordering a couple of each to hack). With just one set, you could use one for the body position reference (There's acceleration sensors in both the remote and the nunchuck AFAIK (but I ain't gonna take my GFs Wii appendixes apart - she would kill me in my sleep  ).