Reproduce “DuoZone: A User-Centric, LLM-Guided Mixed-Initiative XR Window Management System” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jing Qian, George X. Wang, Xiangyu Li, Yunge Wen, Guande Wu, Sonia Castelo Quispe, Fumeng Yang, Claudio Silva
DuoZone introduces a user-centric, LLM-guided mixed-initiative system for managing virtual windows in Extended Reality, combining spatial arrangement zones with AI recommendations. It significantly reduces interaction time and cognitive load for configuring complex workspaces, achieving a 34.5% speed improvement for micro-operations compared to manual methods.
Key insights
- A mixed-initiative human-AI collaboration model, where AI offers an initial 'scaffold' for workspace setup and users retain agency for refinement, proves highly effective for XR window management.
- Integrating spatial anchoring through Arrangement Zones with LLM-driven, context-aware recommendations via Recommendation Zones substantially reduces interaction effort and cognitive load.
- Ensuring both readability of virtual content and maintaining awareness of the physical environment through features like occlusion-free zones is crucial for practical and comfortable XR productivity.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.