Skip to content
Book Open access

Vision: HomeLM for Ambient Intelligence in the Smart Home

Oct 2026 · Proceedings of the 4th International Workshop on Human-Centered Sensing, Modeling, and Intelligent Systems · 0 citations

Abstract

Smart homes hold abundant sensing opportunities, from microphone arrays and presence radars to wearables and everyday device interactions. These modalities rarely add up to a coherent picture of what is happening inside the home. Today's ambient intelligence stitches together many narrow, task-specific models but lacks a unified semantic understanding of the signals they consume. We envision HomeLM, a foundation model that learns to align home-sensing signals with natural language, so that a single model can describe and reason about human activity across a whole house rather than classifying it one task at a time. This would let a home recognize situations it was never explicitly trained for and act proactively on behalf of its occupants. This paper outlines our vision for HomeLM: how such a model could be trained, the research challenges in realizing it, and the opportunities it brings for the smart home.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.