Potential for autonomous linear, curvilinear, and phase detection in natural context para skating using IMU sledge motion data

Abstract

Inertial measurement units (IMUs) were investigated as a means to provide data for sports analytics for para ice-hockey. Algorithms were developed to automatically detect linear skating cycles (i.e., stroke count) and turns. On-ice data are from 2 task expert participants and 5 task naïve participants. Algorithms correctly identified 100% of the trials analyzed, using manually analyzed video recordings as a ground-truth. In addition, visual analysis suggests that IMU data is able to differentiate between task expert and task naïve participants, potentially allowing for performance indicators to be derived. Results from this preliminary study suggest strong potential for IMUs to be a useful tool for sports analytics in para ice-hockey.