Pattern Recognition in Cyclic and Discrete Skills Performance from Inertial Measurement Units

Ludovic Seifert (1) , Maxime L’hermette (1) , John Komar (1) , Dominic Orth (2) , Florian Mell (1) , Pierre Merriaux (3) , Pierre Grenet (4) , Yanis Caritu (5) , Romain Hérault (6, 7, 8, 9) , Vladislavs Dovgalecs (10) , Keith Davids (11, 12)

1 CETAPS - Centre d’études des transformations des activités physiques et sportives
2 VU - Vrije Universiteit Amsterdam [Amsterdam]
3 IRSEEM - Institut de Recherche en Systèmes Electroniques Embarqués
4 MOVEA SA
5 MOVEA SA
6 LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
7 DocApp - LITIS - Equipe Apprentissage
8 INSA Rouen Normandie - Institut national des sciences appliquées Rouen Normandie
9 NU - Normandie Université
10 MODEST
11 Sheffield Hallam University
12 JYU - University of Jyväskylä

The aim of this study is to compare and validate an Inertial Measurement Unit (IMU) relative to an optic system, and to propose methods for pattern recognition to capture behavioural dynamics during sport performance. IMU validation was conducted by comparing the motions of the two arms of a compass, which was equipped with IMUs and reflective landmarks detected by a multi-camera system. Spearman's rank correlation tests showed good correlations between the IMU and multi-camera system, especially when the angles were normalized. Bland-Altman plot, root mean square and the normalized pairwise variability index showed low differences between the two systems, confirming the good accuracy levels of the IMUs. Regarding pattern recognition, joint angle and limb orientation was respectively studied for 25 m during breaststroke swimming and 10 m of indoor rock climbing in athletes of various skill levels. Pattern recognition was also conducted on a macroscopic parameter that captured inter-limb coordination. IMUs revealed the potential to assess movement and coordination variability between and within individuals from joint angle measures in swimming and limb orientation time-series data in climbing.

Mots clés: inertial measurement unit; coordination dynamics; movement variability

Domaines: Machine Learning [stat.ML]

Ludovic Seifert, Maxime L’hermette, John Komar, Dominic Orth, Florian Mell, et al.. Pattern Recognition in Cyclic and Discrete Skills Performance from Inertial Measurement Units. Procedia Engineering, 2014, 72, pp.196-201. ⟨10.1016/j.proeng.2014.06.033⟩. ⟨hal-02096541⟩