Skip to main content

Validation of a Low Cost, Wearable, Inertial Sensor-based Motion Capture System

Studying animal and human positional and locomotor behavior can reveal important information about how bone morphology and biomechanical function are intertwined. Researchers have used optical motion capture systems as a reliable way to capture the kinematics of movement. However, optical systems are usually restricted to a controlled laboratory settings which constrains what positional and locomotor activities can be sampled.  A device with inertial measurement units (IMUs) that can track the kinematics of motion outside of the laboratory setting would provide several advantages. However, IMUs pose problems such as being sensitive to magnetic fields which makes calibration of the device difficult. This paper deals with different ways of calibration and fusing data with two different filters and compares IMU based estimates of joint angles to those obtained from an optical motion capture system.  We do this to determine which calibration and fusing algorithm to use to obtain the best kinematic data. Our team used max/min calibration and offline sphere fitting calibration and applied both the Madgwick and Kalman filter. Our experiments revealed that Offline Ellipsoid Fit calibration improves the accuracy of IMU orientation estimates. They also show that Kalman filter fusion works better than the Madgwick fusion, and that proper orientation estimates for sensors placed on different limb segments rely heavily on our choice of sensitivity settings (frequency scale ranges).

Marisol Guzman
Harvey Mudd College
Research Advisor: 
Dr. John Polk
Department of Research Advisor: 
Year of Publication: