Yusi Sun | Searcharxiv

arXiv2026

JARVIS: A Just-in-Time Augmented Reality VLM-Powered Instruction System for Cross-Reality Task Guidance

Many everyday tasks rely on external tutorials such as manuals and videos, requiring users to constantly switch between reading instructions and performing actions, which disrupts workflow and increases cognitive load. Augmented reality (AR) enables in-situ guidance, while recent advances in large language models (LLMs) and vision-language models (VLMs) make it possible to automatically generate such guidance. However, existing AI-powered AR tutorial systems primarily focus on physical procedural tasks and provide limited support for hybrid physical and virtual workspaces. To address this gap, we conduct a formative study of cross-reality tasks and identify key requirements for state awareness and cross-reality coordination. We present JARVIS, a VLM-driven AR instruction system that generates contextual, step-by-step guidance from a single prompt, with real-time state verification and adaptive visual feedback. To inform the system design, we conducted a formative study to understand guidance needs across cross-reality tasks, which we categorize into four types, real-to-real (R2R), real-to-virtual (R2V), virtual-to-real (V2R), and virtual-to-virtual (V2V). A within-subjects study (N=14) across four domains shows JARVIS improves usability, workload, success rate, and visualization effectiveness over baselines.

cs.HC↗

arXiv2025

Object-Driven Narrative in AR: A Scenario-Metaphor Framework with VLM Integration

Most adaptive AR storytelling systems define environmental semantics using simple object labels and spatial coordinates, limiting narratives to rigid, pre-defined logic. This oversimplification overlooks the contextual significance of object relationships-for example, a wedding ring on a nightstand might suggest marital conflict, yet is treated as just "two objects" in space. To address this, we explored integrating Vision Language Models (VLMs) into AR pipelines. However, several challenges emerged: First, stories generated with simple prompt guidance lacked narrative depth and spatial usage. Second, spatial semantics were underutilized, failing to support meaningful storytelling. Third, pre-generated scripts struggled to align with AR Foundation's object naming and coordinate systems. We propose a scene-driven AR storytelling framework that reimagines environments as active narrative agents, built on three innovations: 1. State-aware object semantics: We decompose object meaning into physical, functional, and metaphorical layers, allowing VLMs to distinguish subtle narrative cues between similar objects. 2. Structured narrative interface: A bidirectional JSON layer maps VLM-generated metaphors to AR anchors, maintaining spatial and semantic coherence. 3. STAM evaluation framework: A three-part experimental design evaluates narrative quality, highlighting both strengths and limitations of VLM-AR integration. Our findings show that the system can generate stories from the environment itself, not just place them on top of it. In user studies, 70% of participants reported seeing real-world objects differently when narratives were grounded in environmental symbolism. By merging VLMs' generative creativity with AR's spatial precision, this framework introduces a novel object-driven storytelling paradigm, transforming passive spaces into active narrative landscapes.

cs.HC↗