Regresar

Angle Estimation of Wrist Movements Through Surface EMG Signals

Abstract:

The present research exposes the wrist position estimation through electromyographic signals, to achieve this, the selection of the angular positions in the movements Flexion- Extension and Abduction-Adduction was made, being the main basis for the development of this investigation. Subsequently, the muscles from which the myoelectric signals were extracted were chosen, according to parameters such as muscle size, superficiality and the contribution to the movement in execution. A study muscle was defined for each movement, electromyographic samples were extracted with the help of electronic cards and surface electrodes of Silver / Silver Chloride (Ag / AgCl), then a database was formed of each movement. A database is formed, the processing, conditioning and digitalization of the signal was carried out; next, the characterization was performed with time domain analysis, obtaining the specific characteristics of the EMG signal. Subsequently, to obtain the main characteristics, machine learning techniques were applied, analyzing the classification with several supervised type algorithms such as Neural Networks, Vector Support Machines, Discriminated Analysis and Nearest Neighbor; this, in order to find an optimal classifier that allows the successful identification of the movement. The research's main claim is to carry out in order to use this methodology in the creation of wrist rehabilitation systems.