Inertia-1 Unified Motion Model Enables Cross-Placement Sensor Transfer
July 20, 2026
Inertia-1 is a foundation model pretrained on wrist data that transfers motion representations to unobserved placements like hips, ankles, and chests. The architecture supports multi-stream sensor fusion and remains robust to sampling rates between 1 Hz and 20 Hz for tasks including gait analysis and activity recognition.
HOW THIS AFFECTS YOU
●
builderYou can deploy a single motion backbone across diverse wearable form factors without retraining for each sensor placement.
●
healthThe model links passive movement data to clinical health markers for long-horizon disease prediction.