Research
Research Agenda
Harmony research branches and future questions
Three interconnected research branches organized around Harmony One. Together they define a working agenda for how spatial intelligence systems might perceive, adapt, and earn trust.
The practical question behind the agenda is simple: how can XR + AI systems preserve context, adapt to changing workflows, and explain their behavior in ways people can trust?
Multimodal Context Inference
How XR systems perceive and represent human state, task progress, and environment through fused multimodal sensing.
Focuses on continuous, non-intrusive observation of physiological, behavioral, and environmental signals — building a live representation of what the user is doing, how they're doing it, and what they need next. Addresses the signal fusion problem that makes adaptive XR testable in deployed settings.
Cognitive Load-Aware Interface Adaptation
How spatial interfaces change layout, complexity, and visibility to match human attention and cognitive load in real time.
Investigates how the interface layer of an XR system can respond intelligently — reducing complexity when load is high, surfacing context when it's needed, and disappearing when it isn't. Builds on Branch 1's inference signals to drive concrete, measurable interface decisions.
Explainable AI Mediation in Spatial Computing
How AI systems in XR share control with humans in a transparent, trustworthy, and user-governed way.
Addresses the trust gap that makes AI-driven XR hard to deploy responsibly. Examines how systems communicate their reasoning, how users maintain agency over adaptive behavior, and how explainability requirements shape architecture from the ground up.
Working framework: Harmony: Adaptive Human-Centered Spatial Intelligence for Context-Aware XR Systems · View research profile ↗