Effect of Garment Attachment Location on the Repeatability of IMU-Based Upper-Limb Motion Measurements

Abstract

Garment-integrated inertial measurement units (IMUs) are widely used for wearable motion sensing, but their measurement repeatability can be affected by relative motion between the garment and the body. This letter investigates how attachment location influences IMU measurement repeatability during repeated upper limb motions. Garment-mounted IMUs were compared with paired body-mounted reference sensors during repeated arm-elevation tasks. Sensors were placed at four upper limb locations—wrist, forearm, upper arm, and shoulder—on a regular-fitting sports jacket. Relative angular deviations were computed using a quaternion-based angular difference metric to compare repeatability across attachment locations. Experimental results from six participants, analyzed using a one-way repeated-measures analysis of variance with attachment location as the within-subject factor, revealed clear location-dependent differences in measurement behavior. The wrist exhibited the lowest trial-to-trial variability, followed by the forearm, whereas the shoulder exhibited the greatest variability despite a smaller mean angular deviation. A significant main effect of location was found for both mean angular deviation [F(3,15) = 7.761, p = 0.002] and trial-to-trial variability [F(3,15) = 10.467, p = 0.001]. These results suggest that attachment location is an important factor in the repeatability of garment-integrated IMU measurements under the present experimental conditions and provide preliminary guidance for sensor placement in wearable motion-sensing systems.