Processing of sEMG signals for online motion of a single robot joint through GMM modelization

This thesis aims to explore the possibility to use EMG to train a GMM in order to estimate in realtime the bending angle of a single human joint. Extraction of features was performed with use of time-domain or frequency-domain transforms and with use of wavelet transform (WT) of which best configuration was investigated. WT was applied to signals of muscles of lower limb. A ROS based software written in C++ capable of interfacing with a humanoid robot was developed