Magnetometer-Based Drift Correction During Rest inIMU Arm Motion Tracking.

Area of Science:

• Human-Computer Interaction
• Robotics and Control Systems
• Biomedical Engineering

Background:

• Real-time human arm motion capture is crucial for applications like video games and physical therapy.
• Off-the-shelf Inertial Measurement Units (IMUs) are commonly used but require homogeneous magnetic fields for accurate orientation tracking, which is often unavailable indoors.
• Existing methods struggle with orientation drift in inhomogeneous magnetic fields.

Purpose of the Study:

• To present a novel method, Rest-Pose Magnetometer-based drift Correction (RPMC), for robust real-time arm motion tracking.
• To address the challenge of orientation drift in environments with inhomogeneous magnetic fields.
• To improve the accuracy and reliability of IMU-based arm tracking.

Main Methods:

• Sensor orientation is estimated by integrating gyroscope angular velocity.
• Drift around pitch and roll axes is corrected using accelerometer data.
• A rest pose calibration records local magnetic field and sensor orientation, allowing magnetometer correction for drift during user rests.

Main Results:

• The RPMC method demonstrated accurate (RMS error = 6.1), low-latency (61 ms) real-time wrist orientation tracking for up to an hour.
• Error reduction compared to alternative methods ranged from 82.5% to 90.7%.
• The method is robust to long-term drift in inhomogeneous magnetic fields.

Conclusions:

• RPMC effectively corrects orientation drift in arm motion capture using IMUs, even in challenging magnetic environments.
• The method leverages the natural need for rest during arm movements to recalibrate.
• This technique enhances the feasibility of reliable arm tracking for various applications in uncontrolled environments.