Estimating Thorax and Shoulder Motion Using Magnetic-Free Quaternion-Based Functional Sensor-To-Segment Calibration.

Area of Science:

• Biomechanics
• Robotics
• Sensor Technology

Background:

• Wearable rehabilitation robots require precise body motion sensing for effective upper-limb therapy.
• Existing 9-axis inertial sensors face challenges in accurately interpreting complex upper-limb movements, often requiring multiple sensor units and magnetometer data.
• Magnetometer data can be unreliable due to ferromagnetic disturbances from actuators near the sensors.

Purpose of the Study:

• To develop a novel methodology for estimating thorax and shoulder motion using only two inertial sensors.
• To overcome limitations of magnetometer use in wearable robotic systems for rehabilitation.
• To provide accurate orientation data for upper-limb rehabilitation robot development.

Main Methods:

• Utilized internal fusion of accelerometer and gyroscope data from two sensors to estimate orientation.
• Implemented a functional sensor-to-segment calibration technique.
• Employed Principal Component Analysis (PCA) on functional movement data to identify primary rotation axes and align sensor/anatomical coordinate systems.

Main Results:

• Achieved accurate tracking of thorax and shoulder motion without magnetometer reliance.
• Experimental evaluation demonstrated a tracking error of 4.07°-5.14° for shoulder orientation.
• Experimental evaluation demonstrated a tracking error of 1.28°-3.88° for thorax orientation when compared to a reference system.

Conclusions:

• The proposed methodology successfully estimates thorax and shoulder motion using two inertial sensors.
• This approach provides a viable alternative to magnetometer-based systems, mitigating ferromagnetic interference.
• The findings support the advancement of upper-limb rehabilitation robots through enhanced motion tracking capabilities.