The Cognition for Technical Systems research group in Munich has been working on improved robot vision systems based on fly vision. Why?

The fly's brain is hardly bigger than a pinhead, too small by far to enable the fly's feats if it functioned exactly the way the human brain does. It must have a simpler and more efficient way of processing images from the eyes into visual perception, and that is a subject of intense interest for robot builders.

The article notes that some flies process images at 100 frames per second, allowing in-flight obstacle avoidance with millisecond response time, something you may have noticed if you ever tried to swat a fly. In particular the researchers are focusing (no pun intended) on optical flow, or as they call it, "optical flux fields". The fly's optical flux field implementation is made up of a first layer of neurons that process the raw input from each compound eye element and feed it to a second stage composed of only 60 neurons for each hemisphere, which reduce the visual field to a series of motion vectors giving the speed and direction of everything it sees. Condylostylus fly macro image by flickr user Opo Terser