Synchronisation of multiple unconnected inertial measurement units using software correction

Abstract

A major challenge in capturing multi-segmental movements with unconnected inertial measurement units (IMUs) is synchronisation between IMUs. The aims of this study were to assess the reproducibility of desynchronisation rates between unconnected IMUs (Axivity, Ax6) commonly used in human movement studies and to determine the accuracy of predicted (corrected) clock differences under different conditions. In the first two experiments, we report that rates of desynchronisation between IMU pairs were linear, unique to each pair, and reproducible within and between sessions. The third experiment involved a cohort of active adults (n = 44) performing physical activity and resulted in predicted clock errors from −10.1 to 0.3 ms after 2 h. This level of synchronisation is acceptable for most human movement applications. The consistent and predictable desynchronisation rates found in these commonly used unconnected IMUs provides an opportunity for a simple, movement-independent, and adaptable techniques to extend synchronisation periods for many applications in human movement research. Further work to compensate for fluctuations in external and internal factors is warranted to extend synchronisation between unconnected IMUs for even longer duration.