Estimating Magnetic Field at Joint Centers Reduces Kinematic Errors in Inertial Motion Capture.

Area of Science:

• Biomechanics and Human Motion Analysis
• Wearable Sensor Technology
• Robotics and Control Systems

Background:

• Inertial measurement units (IMUs) are common for human motion tracking but less accurate than optical motion capture.
• Traditional IMU methods rely on gravity and magnetic north assumptions, often violated by linear accelerations and magnetic distortions.
• Magnetometer-free IMU methods reduce drift but still have limitations in dynamic environments.

Purpose of the Study:

• To develop an improved algorithm for estimating joint angles from IMUs.
• To enhance the accuracy of human motion measurement, particularly for lower extremity kinematics.
• To provide an open-source solution for advanced IMU-based motion analysis.

Main Methods:

• Developed the Magnetic Field at Inertial Joint Center (MAJIC) filter, combining common acceleration and adaptive magnetic field data.
• The MAJIC filter dynamically incorporates magnetic field information to mitigate drift errors.
• Evaluated MAJIC filter accuracy against optical motion capture during ambulation tasks with 11 participants.

Main Results:

• The MAJIC filter achieved a median RMSE of 7.1° for joint angles, outperforming global sensor fusion (9.3°) and magnetometer-free methods (7.5°).
• MAJIC demonstrated a more consistent range of RMSEs across all joints (5.5°–11.4°) compared to other methods.
• The filter showed significant improvements in reducing errors during dynamic human motion tasks.

Conclusions:

• The MAJIC filter significantly enhances the accuracy of IMU-based human motion measurement.
• This method offers a more robust solution for lower extremity kinematics compared to existing techniques.
• The open-source availability of the MAJIC filter aims to advance motion analysis in real-world settings.