Right now I am working on a field within robotics called computer vision. It is exactly what it sounds like. What I am trying to do is to extract useful bits of information from videos. In my case I am looking at extracting three dimensional structural information regarding human kinematics, by having a computer ‘watch’ lots and lots of videos, and then create a huge data-set for me to work on later.
Needless to say this is very difficult, because no two videos are alike.