Yes, we do this to estimate ~sagittal axes from walking rather than relying on people calibrating each day. I like the idea of building in some constraints to improve estimations when uncertainty increases but wonder how this might impact cases where gait truly varies within or between people.

Thoughts on estimating or defining sensor-to-segment position and orientation for real-world uses where direct measurement is impractical? Our current approaches use functional orientation or assumptions about orientation relative to gravity but this limits which biomechanics you can estimate.

Enjoying this thread! "Knowing" the position and orientation of an IMU on a segment is an important step for biomechanics that require estimates of segment orientation. Thanks for sharing how you're measuring or defining this (I sometimes feel that we don't advertise assumptions enough).