Kinect 360 object scanner (WIP, v0.2)
Description
Not working yet, but figured I'd put it up so I could get critiques and suggestions as I work.
The code base is currently in progress, will be borrowing heavily from Kyle McDonald's work for Makerbot, of course. Am sketching it up at the moment, unsure whether I want to do it in Java (very familiar) or Cpp through OF.
v0.2 (files up, not tested or 100%) Defined solution for hub. 4 piece design with bearings.
v0.15 Physical testing and modification of outerJoint and innerJoint. These include adding radial support screws, modified teeth, and refined build methods. The failed code is retained in the build methods with explanation, just for the hell of it.
v0.1 Framework of object and base line openSCAD file. Only parts printed/tested so far are arm joints.
Instructions
Export all export-ready modules.
Attach motor and arduino, upload code.
Hang over object.
Connect Computer (I have a plan for routing the cables, but is still in sketchbook)
Scan.
Print.
You must be logged in to post a comment.
Further updates are on their way. Am on vacation atm and working a bit on grad portfolio. Am expecting to have this project at 1.0 (first release software both embedded and computer, full object, fully tested) finished sometime in December, to give you all a rough timeline. Progress in will be posted and shared, and I continue to appreciate feedback! Feel free to ping me at @readywater on twitter or basically anywhere else.
Major Major Awesomeness!
Any progress on the software end yet?
Wish I could help in some way.
Ya, there's some. I teach kinect coding/design courses in Toronto, and I've some code which I intend to use as a baseline. I want to get a first draft of the object out first, then shift focus to the software base, as development on that side will be a lot quicker. Any suggestions would help, critique and similar would be awesome. I should probably make a git repo for this so people can fork if they like.
How about adding a bubble level under the center in the direction of the arm. Then use a counterweight on a sled you can screw back and forth until you are level and then lock with a couple of nuts. A simplified version of http://www.thingiverse.com/thi... with a hand crank would also do nicely.
Regarding a counterweight, why not weigh the kinect + the portion of the arm that does not get longer, and make a counterweight 4 times as heavy. Hand that counterweight on an arm that is controlled by the same gear mechanism as the extending arm, but with a 4 times reduction. The kinect arm is extended on the side of the gear that has 40 teeth, and the counterweight is on the side that has 10 teeth...
Seems backwards... Why rotate the kinect and not the object?
I'm guessing so that you can scan big objects. I like this design because I've been thinking of doing the same thing for my spinscan laser scanner so I can scan my head instead of figuring out how to make a turntable big enough to rotate my extreme mass. :) Looks like it could be adapted to hold a webcam and a laser instead of a kinect.
The object stays stationary, and the kinect is rotated around it. The spot with the gears is the axis around which it rotates.
License
More from Readywater
This would be the perfect thing to combine with the software from http://reconstructme.net You might want to take a look at that instead of writing your own software. :)
Definitely! Ya, this project has been put slightly on hold while I've been doing my grad applications, but it's coing back before too long! There's been some progress toooo.