Towards Human Capture Movement: Estimation of Anatomical Movements of the Shoulder

In this paper we focus on the human arm motion capture, which is motivated by the requirements in physical rehabilitation and training of stroke patients in the same way as monitoring of elderly person activities. The proposed methodology uses a data fusion of low-cost and low-weight MEMS sensors jointly to an a priori knowledge of the arm anatomy. The main goal is to estimate the arm position, the anatomical movements of the shoulder and its accelerations. We propose a discrete optimization based-approach which aims to search the optimal attitude ambiguity directly without decorrelation of ambiguity, and to computing the baseline vector consequently. The originality of this paper is to apply the discrete optimization to track the desired trajectory of a nonlinear system such as the Human Movement in the presence of uncertainties. The global asymptotic convergence of the nonlinear observer is guaranteed. Extensive tests of the presented methodology with real world data illustrate the effectiveness of the proposed procedure.