Mobile Robot Navigation using Gaze Contingent Dynamic Interface

Abstract

Using eyes as an input modality for different control environments is a great area of interest for enhancing the bandwidth of human machine interaction and providing interaction functions when the use of hands is not possible. Interface design requirements in such implementations are quite different from conventional application areas. Both command-execution and feedback observation tasks may be performed by human eyes simultaneously. In order to control the motion of a mobile robot by operator gaze interaction, gaze contingent regions in the operator interface are used to execute robot movement commands, with different screen areas controlling specific directions. Dwell time is one of the most established techniques to perform an eye-click analogous to a mouse click. But repeated dwell time while switching between gaze-contingent regions and feedback-regions decreases the performance of the application. We have developed a dynamic gaze-contingent interface in which we merge gaze-contingent regions with feedback-regions dynamically. This technique has two advantages: Firstly it improves the overall performance of the system by eliminating repeated dwell time. Secondly it reduces fatigue of the operator by providing a bigger area to fixate in. The operator can monitor feedback with more ease while sending commands at the same time.