Novel Positional Tracking Idea
Posted: Mon Jan 14, 2013 9:39 am
I'm not going to have time to attempt this, so I thought I'd share:
You could do positional tracking using the speckle pattern from diffuse laser light.
Sketch of prototype:
A couple of lasers (of the handheld low power type ~1-5mw), with diffusers pointing inwards into a space.
A couple of webcams mounted to the Rift, with their lenses removed.
The lasers will produce a speckle effect on the CCDs as the coherent light interferes with itself.
You can then use SIFT or other to grab the features from the CCDs and insert into an SFM (Structure from motion) pipeline.
This would be an adaptation of this:
http://specklesense.media.mit.edu/
Naively using SFM on the input is likely to transform in place rotations of the CCD to orbits within the pose estimation space, but I'm sure you can adjust for this.
I couldn't say how well this would work in practise.
You could do positional tracking using the speckle pattern from diffuse laser light.
Sketch of prototype:
A couple of lasers (of the handheld low power type ~1-5mw), with diffusers pointing inwards into a space.
A couple of webcams mounted to the Rift, with their lenses removed.
The lasers will produce a speckle effect on the CCDs as the coherent light interferes with itself.
You can then use SIFT or other to grab the features from the CCDs and insert into an SFM (Structure from motion) pipeline.
This would be an adaptation of this:
http://specklesense.media.mit.edu/
Naively using SFM on the input is likely to transform in place rotations of the CCD to orbits within the pose estimation space, but I'm sure you can adjust for this.
I couldn't say how well this would work in practise.