Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2016
Session Number:P1
Session:
Number:P1-7
Biomimetic Robot Hand Control By Using Surface Electromyography
Stephen Ryan Angsanto, Jonghun Kwon, Jungho Park, Gwanwoo Kim, Wansu Lim ,
pp.857-860
Publication Date:2016/7/10
Online ISSN:2188-5079
Summary:
Undesirable events may happen to people at unexpected times. About 15% of the world's population lives with some form of disability. Most prosthetic devices have limited number of gestures and cannot fully replicate movements of the body. The system utilizes non-invasive surface electromyography (sEMG) for extracting signals from the forearm and uses neural network for pattern recognition. The EMG signals were recorded and calibrated for one participant only. The datasets were sampled to create the input matrix, which are loaded to MATLAB for training, validation, and testing. As shown after successive trials, fatigue or muscle weakness is a significant factor in creating neural networks for pattern recognition. It was verified that the system could successfully extract, classify and output 10 individual finger gestures and 4 manual grasps with a classification accuracy of 93.6%. Statistical analysis was used to assess the classification accuracy based on the results and the original training data with 99% level of confidence.