A visual attention network for a humanoid robot

Abstract
For a humanoid robot to interact easily with a person,the robot should have human-like sensory capabilitiesand attentional mechanisms. Particularly usefulis an active vision head controlled by a visual attentionsystem that selects viewpoints in the environmentas a function of the robot's task. This paper describesa model of human visual attention called FeatureGate,which is a locally connected, pyramidal, artificial neuralnetwork that operates on 2D feature maps of theenvironment....

This publication has 14 references indexed in Scilit: