Regularity assessment of cyclic human movements: An innovative method based on wearable sensors

This study aims at evaluating the regularity of periodic human movements through the analysis of the autocorrelation function of accelerometric signals obtained from wearable sensors. For this purpose a method has been implemented which can run also in real-time, and it has been tested in several experimental conditions and in various motor gestures including gait. Relevant for clinical and sport applications is the fact that the autocorrelation parameters are calculated on the acceleration module, and not on the individual axial components. This allowed us to assess the regularity of periodic movements without the need of aligning the sensor axes to the anatomical axes. Finally a multi-sensor approach has been developed that can be used to simultaneously assess the regularity on different anatomical landmarks.