Sensitivity analysis of human motion for the automatic improvement of gestures

Pauline Maurice (1) , Vincent Padois (1) , Yvan Measson (2) , Philippe Bidaud (1)

1 ISIR - Institut des Systèmes Intelligents et de Robotique
2 LIST (CEA) - Laboratoire d'Intégration des Systèmes et des Technologies

Enhancing the performances of technical gestures is a great concern for human beings, and aims both at improving the operational results and at reducing the associated biomechanical demands. Thanks to the advances in human biomechanics and modeling tools, human performances can be evaluated with more and more details. However, finding the right modifications which improve such performances is still addressed with extensive time-consuming trial and error processes. This paper presents a method for automatically providing recommendations to improve human gestures. An optimization-based whole-body controller is used to dynamically replay human gestures from motion capture data, in order to acquire and evaluate the initial gesture. Virtual human simulations are then run to estimate performance indicators, when the gesture is performed in many different ways, in order to compute sensitivity indices for quantifying the influence of the gesture parameters on the performances. Based on this sensitivity analysis, recommendations for gesture improvement are provided. The whole method is validated on a drilling gesture. The consistency of the replayed motion and the significant increase in the performances of the gesture modified according to the provided recommendations confirm the relevance of the proposed approach.

Mots clés: Sensitivity analysis; Motion capture; Motion replay; Virtual human simulation; Whole-body control; Motion analysis; Ergonomics

Domaines: Robotique [cs.RO]

Pauline Maurice, Vincent Padois, Yvan Measson, Philippe Bidaud. Sensitivity analysis of human motion for the automatic improvement of gestures. 2015. ⟨hal-01221647v1⟩