I understand wanting to use gaze as a UX convenience. Has there been work done on human-robot interaction where a person sees a screen to "see" what the robot sees and can point and click to name/identify objects of interest?