Using visual attention in a Nao humanoid to face the RoboCup any-ball challenge

Abstract

Visual attention is a natural tool which allows animals to locate relevant objects or areas in a given scene, discarding the rest of elements present and thus reducing the amount of information to deal with. In this paper we present the design an implementation of a visual attention mechanism based on a saliency map and its implementation in the Nao humanoid. This control mechanism is applied to solve one of the challenges proposed in the RoboCup competition named ”any-ball”. The results obtained are analysed and future works derived from that analysis are presented