Quick-release function for traveling-nut actuator?

[jbendercp]

Starting to draw up plans for a affordable digitizer for use in prosthetic work in low-income settings. The primary function will be to trace the inner surfaces of plaster casts taken from client's residual limbs, converting them into a 3d-surface for computer-based model rectification and potentially CAD/CAM fabrication.

For those of you not quite familiar with the prosthetic fabrication process, imagine a machine that must digitize the inner surface of a unique, hand-made piece of pottery. My basic design is based upon commercially available designs for prosthetics such as this.

The basic idea is to use mount the cast on a turntable, and use a spring=loaded probe arm that ensures contact with the inner surface that is slowly raised out of the model to create a helical group of points (via rotary encoder data) that can then be converted into a 3D-surface.

My thought is to use a basic traveling-nut lead screw design for the vertical translation of the probe - this should be relatively low-cost, and is "geared" in a sense, such as to reduce drift or frequent deviations from calibration. However, ideally I would like to have a "quick-release" feature that quickly and easily allows the probe to be initially placed in a vertical location, and the re-engaged to start the scan. I've done a bit of research and only found a handful of somewhat related features (such as this). Not sure though if this devices fit the application (they also note not for vertical use). Am I barking up the wrong tree with the leadscrew design if the quick-release is a must-have? Or are there simple ways to overcome this "quick-release" that I am overlooking? Any help would be great.

Thanks,
Jason

 

[tbuelna]

Since what you really want is a 3D surface definition of the person's limb, why not develop a software tool using multiple overlaid images taken from different angles using a cell phone digital camera, that can be used to produce a 3D definition of the limb the prosthetic will be fitted to.

 

[jbendercp]

tbuelna,

Good thoughts. Actually some folks have already moved this direction - the 'Structure Sensor' is a $500 3D digitizer iPad add-on that uses the camera to sense a create a 3D surface. Other companies even in traditional settings as well do exactly what you mention - develop 3d models from pictures of various angles and also limb measurements. I am a bit wary of moving this direction for 2 reasons:

1. In low-resource settings we typically only get one shot (one socket, two at most) to get a good fit, in comparison to traditional settings that often use two rounds of thermoplastic 'check sockets' before finalizing a laminated final socket. Such a picture-based system would probably make an acceptable start for the first check-socket in a traditional setting, but not precise enough for the accuracy that is desired to get a one-and-done socket. The Structure Sensor has not seen the prosthetic use as anticipated for this same reason - doesn't capture all the small curves of the limb well enough to make a high-accuracy model. However, seems the tech may be improving and price dropping in this area - high-end optical scanners which are frequently used in prosthetics in the West have dropped from $20,000+ to under $10,000 in a few short

2. By casting first and then digitizing, this gives some hands on time with the limb, which is desirable to locate and apply compression in tolerant areas, control/position limb angle as well as loose tissue, etc. before even rectifying the model. I see this as a potential way to improve initial fit, thus increasing the amount of "one-and-done" sockets and reducing re-dos.

Just some of my thoughts from a more clinical side, but feel free to disagree, interested to see where the discussion leads.

J

 

[zeusfaber]

Do you need the power transfer that a leadscrew design offers?

Could you attach your carriage to a belt with a driven pulley at the bottom and an idler at the top? Options for resetting the device could then either be to unclamp the carriage from the belt, or to loosen the idler to allow the whole lot to run free.

A.

 

[tygerdawg]

Very interesting project. Given the state of the art of 3D laser scanning, I thought there might be a more "electronic" alternative. In your targeted socio-economic market, I suspect processing speed is not important. For your task, extreme accuracy isn't important either.

A very quick & cursory search led me to this product.

http://www.david-3d.com/en/products/starter-kit-2

Constantly amazed at how technology gallops ahead and what clever engineers can dream up.

Just an idea to ponder. Perhaps this can be adapted somehow (saw the cast in half & scan each part?).

TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering

http://www.bluetechnik.com

 

[jbendercp]

Thanks for your insights, a thoughts regarding some of your comments:

zeusfaber: no, power transmission needs will be pretty minimal ( < 5lb I would guess). The leadscrew does however give me good precision (I'm thinking digitising will be done in/around 2-4mm per slice/rotation, so probably need < .5mm of error/backlash), and it maintains its position even when power/force is not applied. Do you think I could get these two qualities out of a pulley system?

tygerdawg: have definitely been impressed with some of the optical systems available, and have spent a significant amount of time trying to find something that fits the implementation. Unfortunately most the "budget" options (< $1K is desired budget) tend to lack the precision (studies have estimated a radial precision of < 0.5mm is needed), accurate scan of small curves, and ability to filter out slight movements in the residual limb during scanning. Your thought of casting, then scanning a split cast is interesting, although doesn't necessarily solve the precision/small curve issue, and the phrase "semi-automated stitching" on the site worries me a bit about getting an accurate mate.

On a note related to the original concept of the leadscrew design, do you think it possible to use a half (~180 degree) section of a tapped hollow shaft rather than a nut to interface with the screw shaft, which is held in place by spring tension, or would this send my backlash/error through the roof? Just thinking then the user could release spring tension to slide the device for quick placement (would use a smooth guide rail as in CNC use to maintain alignment). Another question is whether this type of piece would even be easily machined?

Keep the thoughts coming, enjoying the discussion.

 

[Compositepro]

Use the head drive assembly from a plotter or inject printer. The steeper motor can be back driven by hand but that should not be necessary as position control is pretty tight and speed range is broad.

 

[mcgyvr]

https://www.fuel-3d.com/usa/product/fuel3d-handheld-3d-scanner-system-108/

$1500

 

[mcgyvr]

http://www.makerbot.com/blog/category/scanner/digitizer

$800

maybe one of those might work... can't beat the pricing now..

 

[jbendercp]

mcgyvr,

Thanks for the thoughts. Not sure either of these would have the accuracy, and/or filtering of limb movement necessary for prosthetic use. Similar issues when people tried to use the Structure Sensor (only $500!!) and even the new Kinect Sensor.

However, I just saw that one of the prosthetic manufacturers in the US has picked up the Structure Sensor and are re-doing the software to correct some of the changes. Looks like my prayers may be being answered as we speak.

Another option, such as tbuelna suggested, may be feasible thanks to a orthotic company that has just last month launched the "Smart Soc" This seems to be a good solution to improve the accuracy of photogrammetry for prosthetic use.

Thanks again, any other thoughts/suggestions are welcome.

J

 

[tygerdawg]

zeusfaber: no, power transmission needs will be pretty minimal ( < 5lb I would guess). The leadscrew does however give me good precision (I'm thinking digitising will be done in/around 2-4mm per slice/rotation, so probably need < .5mm of error/backlash), and it maintains its position even when power/force is not applied. Do you think I could get these two qualities out of a pulley system?

If you need extreme accuracy (I still think you do NOT..."<0.5mm of error" is quite fat) then a ballscrew affair is a good choice. Else I'd use a timing belt / pulley system with proper gear ratios and a small microstepper motor. Relatively low price, easy assembly, good accuracy, minimal or no backlash, very controllable. One of my favorites is Schneider Electric's Intelligent Motion Systems products from

http://www.imshome.com

.

TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering

http://www.bluetechnik.com

 

[RossABQ]

My first thought was a split nut like is used on lathe carriage feeds. A quick quarter-turn of a lever releases the nut from the screw.

https://youtu.be/yqYd2-52R5U