I have a group project idea, A 5 axis 3d digitizing arm

Area for developing group projects, such as workshop equipment
KellyM
Posts: 30
Joined: Fri Dec 31, 2010 10:06 am

I have a group project idea, A 5 axis 3d digitizing arm

Postby KellyM » Mon Jan 24, 2011 11:54 pm

Ok, don't let the 5 axis scare you. What I'm thinking of is an arm that has 5 joints on it(hence the 5 axis) each joint has a sensor to determine relative rotation of the joint.

The sensors could possibly be done with old mouse sensors(the rotational ones, not the laser based ones).

Heres a list of the joints that would be needed, (1)a base rotational and (2)arm elevation, (3)Elbow angle, (4)Wrist Rotational, (5)Wrist Angle.

The sensors could be hooked up to one or more arduino's. (Speed would be the limiting factor, as the sensors from mice are not fully indexed there is a chance of missing a "rotational click" if not fast enough to detect)

The math to locate a point in 3d space is a basic IK(Inverse Kinematic) Chain. I believe it's basic trig functions.

If we could get it down to a plan/kit they could be sold like the makerbot's, etc... and be a source of revenue for the group.

User avatar
otto_pjm
Posts: 750
Joined: Tue Apr 06, 2010 10:12 pm

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby otto_pjm » Tue Jan 25, 2011 10:21 am

It's an interesting thought. I'd be curious if it would yield better results than the line laser and video camera plus turn table solutions out there (Make etc). I've played with those solutions, and the limiting factor is the video resolution.

With an arm I would think the kinematics (knowing where you are based on each arms position) and speed would be the limiting factor. I can tell you from attempting relatively small raster based 3d carving that it can take hours and hours. Also if you are talking about a physical probe you'll likely have to go slower than I carve, unless you have a system that can handle really extreme acceleration. (I.e. driving along at speed and stopping when the probe triggers, before the probe or object being probed is destroyed)

EMC has a open and extensible set of kinematics that might be a good starting point to look at that side of things, with Arduinos I think you be writing them yourself, there of course may be libraries for the maker bot etc that would get you started.

Pete

User avatar
boltz
Posts: 75
Joined: Thu Dec 02, 2010 12:25 pm
Location: Minneapolis

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby boltz » Tue Jan 25, 2011 11:48 am

Pete, are you referring to David for laser scanning?

-Jim Hart

User avatar
otto_pjm
Posts: 750
Joined: Tue Apr 06, 2010 10:12 pm

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby otto_pjm » Tue Jan 25, 2011 11:53 am

I haven't used the David software. I used Grasp. I believe the principle is the same.

I've also seem some web-cam direct to 3D model software demoed, but it was a student project not a software package.

Pete

User avatar
metis
Posts: 1219
Joined: Wed Jan 14, 2009 10:06 am
Location: NE Minneapolis
Contact:

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby metis » Tue Jan 25, 2011 5:58 pm

iirc mouse rotary encoders don't have an index per se, they just have an understanding of + and - so if you get any outside forces that cause the wheel to skip a bit you'll suddenly be way off. it seems like there was a hack using legos adn a b&w disk to produce a pretty solid index though.... (is index the right term here?)

KellyM
Posts: 30
Joined: Fri Dec 31, 2010 10:06 am

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby KellyM » Tue Jan 25, 2011 6:02 pm

I'm actually thinking of something that is not motor driven, but hand driven and not necessarly used for point cloud gathering. but for gathering points in space. ie position the pointer and click a button at which point it gathers the angles from all joints and calculated the point location. or hold the button down and move the arm to gather an array of points as fast as possible.

There's a fairly simple solution to gathering a point cloud allready available in opensource OpenKinect. Which if the resolution is there, may make a 3d digitizing arm unnecessary.

KellyM
Posts: 30
Joined: Fri Dec 31, 2010 10:06 am

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby KellyM » Tue Jan 25, 2011 6:17 pm

metis wrote:iirc mouse rotary encoders don't have an index per se, they just have an understanding of + and - so if you get any outside forces that cause the wheel to skip a bit you'll suddenly be way off. it seems like there was a hack using legos adn a b&w disk to produce a pretty solid index though.... (is index the right term here?)


I think index is the correct term.

I'm not sure what the max rotational rate is for the iirc sensors(I didn't know that's what they were called by the way). Yes, they are only + and - but their should still be a fixed number of steps/rotation of the wheel, I believe the sensors are somewhat of an analog device so as long as the device reading the sensors can handle interupts at a fast enough rate, there should not be lost steps.

I think there are lost steps in mice because the cpu isn't fast enough. This could also be prototyped using hardware counters, I have a Xilinx Spartan 3e FPGA dev board at home.

User avatar
Orkraider
Posts: 657
Joined: Fri Mar 19, 2010 11:14 am
Location: downtown Saint Paul

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby Orkraider » Tue Jan 25, 2011 9:50 pm

it's also possible, if I recall correctly, that "iirc" might stand for "if I recall correctly".
Riley Harrison
facebook.com/rileyharrison

User avatar
boltz
Posts: 75
Joined: Thu Dec 02, 2010 12:25 pm
Location: Minneapolis

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby boltz » Wed Jan 26, 2011 9:08 am

Would it look something like this?
Attachments
screen-shot.jpg
screen-shot.jpg (50.39 KiB) Viewed 1398 times

User avatar
metis
Posts: 1219
Joined: Wed Jan 14, 2009 10:06 am
Location: NE Minneapolis
Contact:

Re: I have a group project idea, A 5 axis 3d digitizing arm

Postby metis » Thu Jan 27, 2011 11:23 am

in my experience lost steps on mice come from dirty wheels, or dirty balls causing contact issues with the wheel drive rollers. since this would be a direct drive (presumably) it's be more precise, however one must remember that you've got a BIG mouse wheel driving a tiny lil roller so that an inch of movement on the rat translates into several revolutions on the wheel. solids gearing would restore that ratio, but w/o great tolerances any take up in the gears will introduce massive amounts of error.

i'm thinking that a lazor mouse as a b&w sensor with a custom printed indexed wheel (so that it can tell from the pattern, and speed of pattern where it is always) will be more precise. you'll still want some bearing or a belt to skew the resolution higher, remember a .01 degree is (iirc) over a mm at a yard out, and once you add in 4 more points of error, and the compound angular skews, you're looking at a significant error in just a tiny movement.

dont' get me wrong, i think this is a brilliant group project, but angles have to be *stupid* accurate to measure with, and i'd rather see us take that into account early on instead of getting things built and discovering that we can't be more accurate than 1/4"

while kinect point clouds aren't bad, the turntable and lazer level and web cam digitizers are apparently quite accurate. what about a 2 axis one of those? mounts could be sort of like lathe ends on rods, or 2 skewers at angles to each other, and then have it turn for a full 360/360 view


Return to “Group Projects”

Who is online

Users browsing this forum: No registered users and 1 guest

cron