Validation of Magneto-Inertial Measurement Units for Upper-Limb Motion Analysis Through an Anthropomorphic Robot

Abstract

Kinematic analysis of human movement is paramount for medical and working scenarios, such as rehabilitation or ergonomic evaluation of workers. Recent studies in the state-of-the-art highlight the importance of assessing the performance of M-IMUs as tools for objective kinematic analysis before exploiting them in any context. This work aims at proposing a method grounded on the use of an anthropomorphic robot, to characterize the behavior of M-IMUs for specific frequency and amplitude ranges, typical of human upper limb movements. The use of a robot allows generating controlled and repeatable angular trajectories, thus providing a gold standard to evaluate the capability of the sensors to accurately track target angular trajectories. The Root Mean Square Error is adopted as a performance indicator. An experimental validation of the proposed method is carried out on two different sets of M-IMUs. The 87.96% and 71.30% of the performed trials exhibit errors lower than 0.087 rad for the WISE and Xsens sensors, respectively. The proposed workflow can be exploited to characterize the sensor behavior for different applications as well as the sensor fusion algorithms implemented into the measurement unit.