Command Selection in Gaze-based See-through Virtual Image-Guided Environments

Embedded close-to-the-eye gaze tracking permits new types of interaction in see-through augmented and virtual environments. It is however unclear how gaze-input can be used to select and confirm commands when the tracking technologies are located at close proximity to user's eyes, but cannot utilize fixed geometry as in screen-based environments. We conducted a study in a simulated image-guided medical environment where users employed gaze-input to control an on-screen display. The current hand-based interaction of such views is a frequent source of interruption and thus feasibility of alternative input modalities has to be evaluated. We created a three-stage gaze-based confirmation mechanism and evaluated its robustness and the limits of the target size. Two sizes of the target for command selection were evaluated, occupying 12 and 6 degrees of visual angle at the 30cm distance. The results show the time to perform an action using gaze input is shorter than in hand-based interaction with the real-world device, confirming that this input modality is feasible. The size of target has little effect on the interaction and the completion-error is low. The findings have implications on the design of future gaze-based input methods for these devices.