Multi-View Web Interfaces in Augmented Reality

Abstract

The emergence of augmented reality (AR) is reshaping how people can observe and interact with their physical world and digital content. Virtual instructions provided by see-through AR can greatly enhance the efficiency and accuracy of physical tasks, but the cost of content authoring in previous research calls for more utilization of legacy information in AR. Web information is a great source hosting a wide range of legacy and instructional resources, yet current web browsing experience in AR headsets has not exploited the advantage of 3D immersive space mixing the real and virtual environments.

Instead of creating new AR content or transforming from legacy resources, this research investigates how to better present web interfaces in AR headsets, especially in a physical task instruction context. A new approach multi-view AR web interfaces is proposed, which suggests separating web components into multiple panels that can be freely arranged in the user's surrounding 3D space. The separation and arrangement would allow more flexible combination of web content from multiple sources and with other AR applications in the user's field of view.

This thesis presents a remote and self-guided elicitation user study with 15 participants that derives layout arrangement preferences of the proposed multi-view interfaces. The study uses a VR system developed to simulate three scenarios of performing real-world tasks instructed by multi-view AR web content involving different types of media. The study analyzes how users arrange such web interfaces, and the system also simulates various physical environments and general AR applications to investigate their impact on the virtual interface arrangement.

According to participant survey responses and interface arrangement data, the study identifies patterns in interface layout, grouping relationships between interfaces, physical environment constraints, and relationships between web interfaces and general applications. Then five implementation strategies are suggested based on the design preference findings.