Validation of functional calibration methods for inertial motion capture units in disturbed magnetic fields

Abstract

In the field of biomechanics, Inertial Measurement Units play an increasingly important role in the kinematic analysis of human movement. The sensors are often instrumented with magnetometers, which give information about the alignment in an external reference frame. Since many users cannot guarantee magnetically homogeneous conditions in their facilities, the use of inertial motion capture technology is hereby limited. Researches and engineers face the challenge of developing easy-to-use calibration methods, which rely on the information of gyroscopes and accelerometers, to overcome this restriction. In this study, two of those calibration methods are described. The procedures, developed by Noraxon Inc, follow different functional, two step approaches. The methods were validated against an optical reference system for movements of the lower limbs in gait and squat. Different data sets with initial pose calibrations in an assumed magnetically homogeneous and in an assumed magnetically inhomogeneous spot were evaluated. Both examined Functional Calibration Methods show promising results for movements in sagittal and coronal plane. The methods correct heavily spoiled data in the inhomogeneous spot and increase Range of Motion and Root Mean Square Errors (RMSE) relative to the optical reference. The RMSE of sagittal movements stayed below 3° for all angles. Movements in the coronal plane show the highest errors for the original data in the disturbed magnetic spot and can be reduced to moderate RMSE below 5°. Qualitative analyzes of gait cycles confirm the findings. The results demonstrate the increased usability of the system, since they are comparable to common protocols for gait analysis. In the transversal plane, IMU data shows higher deviations to the reference up to RMSE of 7.96°. The deviations to the reference were high for both calibration spots, which shows the limitations of IMU technology for movements in this plane. The proposed Functional Calibration methods do not fundamentally increase nor deteriorate the quality of the signal in this plane. Further investigation should focus on analysis of movements in the transversal plane and the offset between the IMU data of different calibration methods.