Speed based classification of mechanomyogram using fuzzy logic

Mechanomyogram (MMG) signals are the mechanical signals obtained from muscles during contractions. They are less sensitive to skin impedance, sensor placement and require only low cost hardware to process the signal. Till date there are only very few applications in which MMG signals are used. The work aims at development of a standalone system for generating control signals required to drive assistive devices which provide support for disabled and elderly people. This paper presents the initial phase of the work, which focuses on the development of a fuzzy classifier. The classifier is developed to categorize the different speeds of elbow movements into rest, slow and fast. For this, MMG signal from biceps brachii are acquired and processed. Two time-domain features namely, mean absolute value and variance are extorted from the segmented data and is given to the fuzzy inference system. The average accuracy of the classifier is found to be 72.72%.