Gaze control using human eye movements

Abstract

In this paper, we propose a visual servoing of the orientation of a camera mounted on a robot, using human eye movements. The study aims at the remote survey of sensitive sites or the intervention in contaminated nuclear sites. Visual data stemmed from an eye-tracker, providing the direction of the look of an operator, is generally noisy and sometimes inconsistent. Filtering, smoothing and stabilizing these eye tracker data are thus necessary in order to move the camera in the direction specified by the operator's eye. The visual servoing control scheme, as well as experimental results, are also presented.