Comparing inertial measurement units and marker-based biomechanical models during dynamic rotation of the torso

Abstract

Inertial measurement units (IMUs) enable human movements to be captured in the field and are being used increasingly in high performance sport. One key metric that can be derived from IMUs are relative angles of body segments which are important for monitoring form in many sports. The purpose of this study was to (a) examine the validity of relative angles derived from IMUs placed on the torso and pelvis; and (b) determine optimal positioning for torso mounted sensors such that the IMU relative angles match closely with gold standard torso–pelvis and thorax–pelvis relative angle data derived from an optoelectronic camera system. Seventeen adult participants undertook a variety of motion tasks. Four IMUs were positioned on the torso and one was positioned on the pelvis between the posterior superior iliac spines. Reflective markers were positioned around each IMU and over torso and pelvis landmarks. Results showed that the IMUs are valid with the root mean square errors expressed as a percentage of the angle range (RMSE%) ranging between 1% and 7%. Comparison between the IMU relative angles and the torso–pelvis and thorax–pelvis relative angles showed there were moderate to large differences with RMSE% values ranging between 4% and 57%. IMUs are highly accurate at measuring orientation data; however, further work is needed to optimise positioning and modelling approaches so IMU relative angles align more closely with relative angles derived using traditional motion capture methods.