A biologically-inspired robotic vision system for tracking fast moving objects

We present a robotic vision system that is capable of tracking fast moving objects. We briefly describe the design and construction of the system and concentrate on the biological strategies we have implemented to improve the system´s capabilities in terms of speed and accuracy. We have also explored new horizons in the area of artificial intelligence to propose some useful algorithms based on visual attention to scan the scene efficiently, and govern the system´s behavior. The algorithm allows us to select regions of interest over which we force the fixations of the vision mechanism. Saliency maps are then created with characteristics, such as color, geometry, optical flow and intensity that highlight within the scene. The amount of maps that can be extracted in an image is huge, so we just use some of them to avoid high latencies that can harm the performance of our system. We use the Winner Take All algorithm to select the most salient region. We have evaluated the performance of our system on real environments.