Robust Plug-and-Play Joint Axis Estimation Using Inertial Sensors

Area of Science:

• Biomechanics
• Robotics
• Sensor Technology

Background:

• Accurate sensor-to-segment calibration is crucial for inertial motion capture.
• Existing methods require specific, informative motion within initial time windows and lack success indication.
• Hinge joint axis identification is essential for applications like human movement analysis and robotics.

Purpose of the Study:

• To develop a plug-and-play calibration method for inertial motion capture.
• To enable accurate joint axis estimation from arbitrary motion data, overcoming limitations of current techniques.
• To provide users with a clear indication of calibration success and accuracy.

Main Methods:

• A novel method combining acceleration and angular rate data for globally optimal joint axis estimation.
• Advanced sample selection techniques to identify informative data from larger datasets, discarding redundant measurements.
• Uncertainty quantification to validate the accuracy and reliability of the estimated joint axis parameters.

Main Results:

• Achieved angular errors of approximately 2 degrees using a selected subset of 125-1000 samples.
• Demonstrated successful calibration using a mechanical joint across a wide range of motions.
• Validated the method's ability to extract informative data even with non-ideal user movements.

Conclusions:

• The proposed method offers the first truly plug-and-play solution for inertial motion capture joint axis calibration.
• It eliminates the need for dedicated calibration phases, providing accurate estimates rapidly.
• The method enhances usability by not requiring specific user motions and assuring calibration validity.