| Real Time-Spatial SEMG Classification for Motion Control Using Fuzzy Inference System
Thesis Abstract--Idaho State University (2018)
Control of prosthetic devices using surface Electromyography (sEMG) signals is a common approach to provide enhanced functionality to upper body amputees. Current sEMG based upper body prosthetic devices are limited by the amount of information drawn from the acquired sEMG signals. The goal of this thesis work is to investigate the use of spatial features along with the temporal variations for the development of a fuzzy logic inference system. The spatial features are deduced by utilizing a circular sensor arrangement. In particular, the sEMG data along with its geometrical position is correlated with human forearm motion. This setup is used to construct fuzzy rules for a simple inference system to predict human hand motion from the sensed sEMG data and its position on the forearm. Five sEMG channels are equally distributed spatially over the perimeter of the forearm, and a polar coordinate system is used to for visualization of the changing sEMG signals. Both the area of the resulting sEMG pentagon and the corresponding inner angles of the pentagon are studied with using the Fast Fourier Transfer (FFT) and Standard Deviation (STD) classification method. The results of the frequency analysis of the sEMG-area property indicates some promise whereas angle based representation of the sEMG data shows to be a good source fuzzy logic inference rule development. The STD of internal angles give distinct relations, which are used to construct a fuzzy logic controller that is capable of discerning with some reliability two forearm movements.
Key Words: sEMG, Spatial features, Fuzzy Logic Controller |