Mach2 Faro Arm Inverse Kinematics

[vacpress]

Here is what i was gonna say, cut n pasted.

For interested people - the subject is a "faro arm" type digitizer that uses mouse encoders, simple materials, and mach2 or similar coftware to generate the "point cloud" file.

Anyone know if Mach2 can recieve 6 axis locations and do the inverse kinematics to create the x,y,z point file?

High Seas wrote on 04-30-2004 12:30 AM: good thanks - I'll check with Art - the MACH2 guy - re those questions - good thinking with the mice - - have a good one! Cheers - Jim

if you keep digging up info on the thing, maybe i will get bored and do a 2quick 3D model of a simple device, that can be made with... whatever... any strong 1/2" thick material like acrylic or aluminum, or some such thing.. maybe tube... and 3 mice..

i bet a pic could be used to decode the mouse signals from the things, so the mice PCBs can be used as well designed amplifiers.

If mach2 dosent provide the needed input code, im sure some hacker here could whip it up. a guy from australia wrote a printer driver for my painter project, which has less universal appeal than a cheap, simple, effetive digitizer arm.

the mouse mechanics would not be used, just the optical encoder & wheels, and the PCB. the ball and the shell would not be used at all..

This conversation should be moved to the forum, as other people may be interested.

[ESjaavik]

To keep the "knuckles" from being too clumsy, you could put the encoder wheel between the bearings in the knucke and inside the arm that moves relative to the knuckle axle goes the encoder fork. Don't forget to put in an extra fork and hole as an index mark! A mouse only give relative movement, but a digitizer arm needs not only know which direction and how far, but also where it started. Just drill a tiny hole in the indexer wheel and put in a fork to read the hole.

[Splint]

Just borrowing an extract from the mach2.pdf available from www.artofcnc.ca

If the mechanism of your machine is like a robot arm or hexapod then Mach2 will not be able control it because of the kinematic caculations that would be needed to relate the "tool" position in x,y and z coordinates to the length and rotation of the machine arms.

So from what I can gather Mach2 can handle a digitizer mounted on the z axis of a 3 axis mill but not an arm type of arrangement or a four or five axis digitizer where the 4th and 5th axies are on the spindle.
If your main interest is point cloud creation as opposed to finding edges and referece points on the stock prior to machining you might want to look at your cad software to see if that could handle a home made digitizer, this of course would almost certainly be subject to a driver file being created so that the cad program can correctly interpret the position of the probe. There may even be someone here on this board who would be willing to help in this area. Also have a look at the yahoogroup cadcamedmdro there may be dome good info there. Dont give up on this project, it sounds like a great idea.
Cheers
Splint

[vacpress]

it seems like a 2-4" probe would be easy to create. i have considered it. when i have a mill up.....

[ger21]

With this one,
http://www.immersion.com/digitizer/

You can do it right in Solidworks, or any one of about 20 other packages.

[Splint]

The Microscribe G2 certainly looks like a quality turn key product but at A$6240.00 it's not really in the realm of an alternative to a home made digitzer. A good basis for a design concept none the less.
The basic principal of how a digitizer works is very simple and I'm sure with some of the talented people on this board the construction of a low cost digitizer with reasonable accuracy and compatability with the major cadcam software is quiet achievable. It'll be interesting to see what becomes of this.
Cheers
Splint

[ger21]

I thought the discussion was about getting data from a Microscribe or similar into Mach2. I didn't realize it you were talking about building a Microscribe-type device. It seems like it would be cheaper and easier to use a standard probe, unless you wanted 4 or 5 axis type data.
It looks like you would need a custom app to read the encoders and write data to a file. Probably not that difficult?

[Cliff_J]

Wow, so many forums, so little time. I posted in the wrong one I guess.
Digitizing arm options

We're on the same page in terms of building a low-cost digitizer using cheap encoders. I think the big trick is first to get the encoders to a precision step and then finding the math to calculate the positions. Some quick math I did says that a 12" arm would create a 75.4" circumfrence and with a 2000 count encoder that puts its position within .0376" and with a second 12" arm we're talking .151" and without better precision it doesn't make sense to find the cumulative error for even more axis. If we can go to a 2048CPR with quadrature to get 8192PPR then I figured .009" for each axis and .074" for 3-axis which seems like the best compromise to minimize part count but offer a somewhat usable arm. My math might be completely wrong, I'm just guessing I understand how this works.

Cliff