Estimation of angle based on EMG using ANFIS

Publication Type:
Conference Proceeding
Citation:
2016 IEEE Symposium Series on Computational Intelligence, SSCI 2016, 2017
Issue Date:
2017-02-09
Full metadata record
Files in This Item:
Filename Description Size
07850065.pdfPublished version1.72 MB
Adobe PDF
© 2016 IEEE. There are wide verities of human movement possible that involves a range from the gait of the physically handicapped, the lifting of a load by a factory worker to the performance of a superior athlete. Output of the movement can be described by a large number of kinematic variables. Modeling each case with a muscle model is difficult. Intended action data can also be extracted from surface Electromyography (EMG) signal which may include intended torque, angle and impedance parameters of the knee joint dynamics. In this paper, Adaptive Neuro-Fuzzy Inference System (ANFIS) has been used in trying to estimate angle. As EMG signal is a function of angle, velocity and muscle activation level (load lifted), an adaptive machine learning technique is most desirable. Many different EMG signal intensity is possible at the same extension angle for different velocity of lower limb movement about knee joint. The EMG signal has been extracted from two different muscles and their patterns are very unique from velocity to velocity for entire range of extension angle. So a learning method of a Neural structure whose connections are based on rules is required to be able to estimate the angle at various speed about the knee joint as the slope of EMG signal intensity for each case of velocity varies significantly. The EMG signal has been collected from volunteer who has completed the knee joint extension in 15 Sec, 10 Sec, 8 Sec, 5 Sec, 3 Sec, 1 Sec, 0.5 Sec and 0.35 Sec respectively. RMS feature has been used to smooth the raw EMG signal. ANFIS is able to estimate angle adaptively although EMG pattern is changing with respect to speed. The simulation has shown experiment of comparative performance of angle estimation by different membership function and features.
Please use this identifier to cite or link to this item: