From Mocap data to inertial data through a biomechanical model to classify countermeasure exercises performed on ISS

Abstract— On board the International Space Station (ISS) resistive training is essential to reduce the effects of musculoskeletal system deconditioning due to weightlessness. However, it could be equally dangerous or not useful if performed with inappropriate techniques. Thus, a system based on inertial sensors able to monitor astronauts has been thought. In this work, an OpenSim biomechanical model was used to reproduce motion of countermeasure target exercises and to simulate inertial sensors put on the model. This was done starting from kinematic data collected with motion capture system (mocap), because no inertial data were available. Then, it was explored a possible approach to build the classifier able to automatically recognize ‘correct’ and ‘wrong’ techniques of execution. Two machine learning algorithms were compared and results in terms of accuracy were encouraging.