So I’m listening to Faithless and thinking I should do a blog entry…

I made a bit more progress with the vision system last night. I think I’ve worked out how to calculate the 3D position of a feature from the intrinsic camera properties and the feature location disparity between the stereo images. I haven’t tested it yet, but if it works I’ll be pretty pleased.

Of course I’ve still got some pretty major challenges ahead of me such as estimating camera ego motion by matching features between frames….

I need to start updating my blog a little more often. I finally started making some real progress on my vision system this weekend. I have been working on this project (in various forms and incarnations) for over a year now so it’s pretty neat to finally start getting somewhere.

I can now extract Scale Invariant features from images and match feature sets between stereo images. This doesn’t sound like much, but it is the first very important step to making my SLAM (Simultanious Localisation And Mapping) vision system a reality!

A huge thanks to Sebastian Nowozin whose libsift project and emails really helped get me off the ground.

Oh and the Egypt photos I posted are from my recent holiday there. I’ll eventually get round to putting up a proper page on my website.