Gaze Interface: Utilizing Human Predictive Gaze Movements for Controlling a HBS

摘要

We explore how gaze can be proactively used as part of the control interface for a hybrid bionic system (HBS) in goal directed tasks. Since human gaze behavior has been shown to support hand movement planning, tracking gaze fixation while doing simple, well-learned, object manipulation tasks provides a natural way for inferring the subjectpsilas motion intent. We devise a simple algorithm based on the gaze fixation area and gaze velocity for sending commands to a robot simulating a HBS. This gaze interface is shown to provide early specification of sequential movement goals according to the subjectpsilas action plan (intention) and early triggering of appropriate movement commands.