In situ sensor-to-segment calibration for whole body motion capture

Abstract

Sensor-to-segment calibration is a critical step for motion reconstruction from inertial and magnetic measurement units (IMMUs). In this paper, a novel sensor-to-segment calibration protocol is proposed. The protocol consists of three stages that allow for in situ calibration. After the sensor units are attached to the body, predefined postures and movements are used for sensor calibration. Acceleration and angular velocity measurements are used to estimate axes of functional frame (FF) by Principal Component Analysis (PCA). Finally, Levenberg-Marquardt optimization is used to identify rotation matrices between the expected FF and their estimations with respect to the sensor frame. Validation of the method demonstrates its practical value and how the proposed protocol reduces the extent of cross-talk for evaluating joint kinematics.