Wii Hack and Mods

Have you guys seen this?

[www.cs.cmu.edu]

Could you use a Wii remote and a few IR points to improve accuracy on a reprap?

Squintz,

Your link is dead. At least, I can't get it to load a page. Interested though.

Demented

It worked a day or two ago, I think that web server is down right now. Maybe CMU's tech staff started their New Year partying already? :-) Even the top level [www.cs.cmu.edu] [http://www.cs.cmu.edu/] is unreachable at the moment.

His stuff about IR pens and a cheap electronic whiteboard was interesting, though it looked like a lot of his work was very Windows-only (less interesting for me, Linux and FreeBSD-using free software nut that I am!).

Jonathan

---

---

Jonathan

jmarsden Wrote:
-------------------------------------------------------
> His stuff about IR pens and a cheap electronic
> whiteboard was interesting, though it looked like
> a lot of his work was very Windows-only (less
> interesting for me, Linux and FreeBSD-using free
> software nut that I am!).
>
> Jonathan

The topic at our upcoming Feburary Kansas City Robotics Society meeting will be about using the Wiimote devices as sensors. One of our presenters is developing a Linux version of the Wiimote Library. I believe he is planning on publishing his work. I'll see what I can find out. Here is a link to our meeting information if you should happen to be in Kansas City on Feb 2.
[www.kansascityrobotics.org]

Vince

If I am remembering the Wii-mote specs correctly, the resolution is 640x480 for the camera on the front and it needs more than one pixel of IR source to be identified as a tracking point, seriously cutting down the available resolution. Which I wouldn't have thought it would give any kind of real accuracy for positioning, although I'd expect it'd be good enough to get a rough position for something.

Taking this further, using a real high resolution camera (8 Mega pixels) with an IR filter on the front with maybe an IR laser source pointing at the bed, may give enough accuracy to get closer for positioning.

It is possible to get sub pixel accuracy when processing video images. Suppose you have a point source of light which activates several neighboring pixels with a peak intensity in the middle. You can do some maths on the intensity of each neighboring pixel that can identify the actual peak to an accuracy which can be greater that the pixel resolution. I.e. the luminance resolution adds to the spacial resolution.

---

[www.hydraraptor.blogspot.com]

Yeah, it's totally possible, it's kind of what 3d laser scanners do, except they use lines. Just doing some quick math, if you have a totally unobstructed view from above of a work area 30cmx30cm, you could use a 300x300 pixel camera to get approx 1mm precision (maybe) and 600x600 pixel to approach 0.1mm.

Although in the real world, we'd have to place the camera off to one corner so close up our precision would be better and less so further away due to distance. Rethinking my earlier post, now that I did based on zero math, going that high on the camera resolution isn't needed now that I took some time and thought about it. I'd say it's quite viable maybe with a 2megapixel camera.

I also thought about using some camera. But what about distortion? I would expect the pixels on the sensor chip to be aligned in a pretty accurate rectangular pattern. But the lens is just a set of glass blobs converting straight lines to curves.

A colleague told me about a former project where they used images taken with such cameras for geographic mapping and they had to calibrate individually for each camera.

Howie

---

--
Airspace V - international hangar flying!
[www.airspace-v.com] for tools and toys

Hi Howie,

... for images with very low distortions you need telecentric cameras, whats pretty expensive, or inverted video-makro-optics, as in a special video-microscope, i developed for microassembly-imaging.

Look at the images here (on the right side): - [www.mikroskopkamera.de] - or here: - [www.mikroskopkamera.de]

The images have nearly no distortion and a very high area of sharpness ...

Viktor

Edited 1 time(s). Last edit at 01/22/2008 07:03AM by Viktor Dirks.

> ... for images with very low distortions you need
> telecentric cameras, whats pretty expensive, or
...
> The images have nearly no distortion and a very


Umm, yes. You're right. Very pretty images! Actually, with 'distortion' I meant bendings due to irregular shapes of the lens. E.g. if the front and back shapes are a bit excentric or if a lens isn't exactly mounted in optical axis... or whatever production imprecision might occur.

If you use a camera for a flat surface a regular non-telecentric lens is sufficient. However, I like the idea.

Howie

---

--
Airspace V - international hangar flying!
[www.airspace-v.com] for tools and toys

I'vee seen an interesting comment this morning on the blog of Johnny Chung Lee - [procrastineering.blogspot.com] - regarding scanning of 3D objects:

[www.blogger.com]

Cambridge university demo of 3D-tracking system:

[www.youtube.com]

"I think the resolution might suffer a bit though, but I don't really know. It might also be a good idea to make an automated scanner that follows a grid; I don't think a projected grid would work since it can only track four points."