Inertia-1: One Unified Motion Foundation Model for Every Body and Task

Inertia-1: An Open Exploration to a Unified Motion Foundation Model

I present Inertia-1, a universal motion model that unifies fragmented datasets into a single backbone. Trained on over 18 million hours of global accelerometry, this system transfers seamlessly across body placements, sensor types, and tasks without retraining. From activity recognition to health monitoring, one representation now adapts to any setting, proving that motion understanding can be truly general.

"Pretrain once on the wrist, then point the model anywhere. It holds up on body placements and even sensor types like gyroscope and magnetometer that it never saw during training."