Entropy-based gaze planning

This paper describes an algorithm for recognizing known objects in an unstr uctured environment (e.g. landmarks) from measurements acquired with a sing le monochrome television camera mounted on a mobile observer. The approach is based on the concept of an entropy map, which is used to guide the mobil e observer along an optimal trajectory that minimizes the ambiguity of reco gnition as well as the amount of data that must be gathered. Recognition it self is based on the optical flow signatures that result from the camera mo tion signatures that are inherently ambiguous due to the confounding of mot ion, structure and imaging parameters. We show how gaze planning partially alleviates this problem by generating trajectories that maximize discrimina bility. A sequential Bayes approach is used to handle the remaining ambigui ty by accumulating evidence for different object hypotheses over time until a clear assertion can be made. Results from an experimental recognition sy stem using a gantry-mounted television camera are presented to show the eff ectiveness of the algorithm on a large class of common objects. (C) 2001 El sevier Science B.V. All rights reserved.