Boosting Vision through Embodiment and Situatedness

Abstract

In active vision the percepts can be acquired in a purposive way by action of a camera. However, it is often necessary to change features of an environment in order to solve a visual task. Some of the features that may be modified by a robotic agent (using for example a robotic arm) are the edges/corners contained in the image (spatial content), the motion (the optical flow computed from the image), the frequency content, the color content, or even the structure of a scene in an image.