Development of a Support System for the Physically Handicapped Using EMG-controlled Interface

点击翻译按钮获取中文摘要

The purpose of this research is to develop a new support system consisting of an interface device and a human assisting manipulator for the physically handicapped. The obtained results throughout this research can be summarized as follows:(1) A new neural network called Recurrent Log-Linearized Gaussian Mixture Network (R-LLGMN) was proposed to realize an adaptive learning ability to a variety of EMG signals measured from the handicapped. The R-LLGMN was developed on the basis of the Gaussian mixture model and the hidden Markov model, and it was shown from experiments that the R-LLGMN has higher pattern recognition abilities than conventional neural networks.(2) Using R-LLGMN, an algorithm to recognize motions of forearm and hand from measured EMG signals of operators was developed to realize the EMG-based interface.(3) A new robot control system manipulated through EMG signals measured from the operator was then developed, and incorporated into a human assisting manipulator. Also, to use it in daily activities, specific knowledge concerning about each task is modeled and incorporated into the control system of the manipulator.(4) As an example of application of the developed EMG-controlled interface, a new robot system for meal assistance is developed. From experimental results by the physically handicapped, a possibility to come into practical use was shown.