Development of an HD-sEMG System With Source Separation Algorithm

To replicate the fine movements of the human hand and to reproduce the synergistic interactions of the muscles, a high degree of freedom telerobotic or prosthetic hand is required. For the detailed control it is essential to detect the human intent with similar detail, even distinguishing the activation of individual muscles. In our lab, we are developing a multimodal control system, including ultrasound-based and near-infrared spectroscopy-based measurements in addition to high density surface electromyography (HD-sEMG). This manuscript focuses on the design of the HD-sEMG system and on the related source separation algorithm. While HD-sEMG solutions are already on the market, our motivation is twofold: to create more flexible compatibility with the other modalities and to extend the empirical limitations related to sEMG known in the literature. In our work, separation between surface and deep muscles of the forearm is reached in non-isometric cases, while the identification of individual muscle groups needs further investigations.