Abstract:
Natural upper limb motion is characterized by efficient coordination between the hand and the arm. Based on underlying coordination patterns, high-density (HD) surface electromyographic (sEMG) signals from the upper limb can be used along with regular sEMG signals to predict hand kinematics. Accurate predictions of hand kinematics from EMG signals are integral for the development of high-performance robotic finger prostheses or increasing the reliability of existing ones. To obtain a better understanding of the mapping between regular and HD-sEMG signals and the kinematics of the upper limb, we created and optimized a novel experimental setup, using which, two pilot experiments were performed on a healthy subject. In these experiments, we measured motion capture data as well as regular and HD-sEMG signals from the subject's upper limb, during directional point-to-point grasping of target blocks located in a fixed plane. We used the Fourier-based Anechoic Demixing Algorithm (FADA) to identify and characterize the spatiotemporal synergies present in the sEMG data. We found the 1:1 mapping between the arm and hand synergies, and the synergies extracted from all the muscles combined. The similarity between the hand synergies and the corresponding synergies from all the muscles combined was much higher than that between the arm synergies and the latter. 3 spatiotemporal synergies were sufficient to successfully reconstruct the 9 directional grasping movements. The reconstructions obtained explained more than 80% of the variation in the data.
