Glance-Box: Multi-LOD Glanceable Interfaces for Machine Shop Guidance in Augmented Reality using Blink and Hand Interaction

Glanceable User Interfaces for Augmented Reality (AR) reveal virtual content “at a glance,” providing rapid information retrieval, often based on gaze interaction. They are ideal when the augmented content covers a small proportion of the view space. When the size of virtual content grows, the potential to occlude the real-world increases provoking safety concerns. Compounding this is the Midas Touch Problem, where users unintentionally select virtual elements by simply looking at them. Extending dwell time does not eliminate involuntary selections, impeding interaction time. In this work, we present Glance-Box, a novel interaction system for AR combining Glanceable interfaces and world-based $3\mathrm{D}$ interfaces across three Levels-Of-Detail, including progressively more information and visuals. Glance-box combines eye-gaze and hand interactions, focusing on user safety. A $2\mathrm{D}$ Glanceable interface facilitates rapid information retrieval at a glance, while extended $3\mathrm{D}$ interfaces provide interaction with denser content and $3\mathrm{D}$ objects. Glance-Box couples blink-based and gaze-based interactions to minimize errors arising from the Midas Touch Problem. While applicable across domains, the Glance-Box interface is designed and optimized for performing manufacturing tasks in the real world. We evaluated the Glance-Box interface using an object selection task of a manufacturing process. Participants completed tasks faster using Glance-Box, employing less dense LOD over time as they gained experience. The perceived accuracy of Glance-Box gaze-based input was high, even when the device's eye tracker accuracy was coarse.