ON THE DEVELOPMENT OF MYOELECTRIC PROSTHETIC HAND CONTROL SYSTEM

ON THE DEVELOPMENT OF MYOELECTRIC PROSTHETIC HAND CONTROL SYSTEM

conference paper

en

This article deals with development of a myoelectric prosthetic hand, that is developed at The Institute of Production Machines, Systems, and Robotics, FME, Brno University of Technology. A computer model of the prosthetic hand was proposed, first mechanical prototype made, and computer modelled mechanical characteristics verified on this mechanical prototype. The mechanical prototype was found applicable for serial production and was presented on The 47th International Engineering Fair at Brno, Czech Republic 2005. Nowadays development of a electronic control system for this prosthetic hand has been started. The prosthetic hand will be controlled through the use of electromyographic signal (EMG) that occurs during a muscle contraction. It is proven that four basic grasp shapes can be recognized from a two-channel EMG signal. The EMG signal is measured through the use of surface electrode. The EMG signal must be preprocessed, which means the filtration, amplification, and signal reconstruction. The EMG signal in original time domain contains huge amount of data and therefore it is not suitable for classification purposes. The EMG signal is processed in a digital signal processor (DSP) where the feature vector by the means of suitable transformation from signal in original time domain is obtained. For signal classification neural networks and artificial intelligence algorithms are used. The actual state of myoelectric prosthesis hand development, description of the electronic control system architecture, and summary of obtained results are described.

This article deals with development of a myoelectric prosthetic hand, that is developed at The Institute of Production Machines, Systems, and Robotics, FME, Brno University of Technology. A computer model of the prosthetic hand was proposed, first mechanical prototype made, and computer modelled mechanical characteristics verified on this mechanical prototype. The mechanical prototype was found applicable for serial production and was presented on The 47th International Engineering Fair at Brno, Czech Republic 2005. Nowadays development of a electronic control system for this prosthetic hand has been started. The prosthetic hand will be controlled through the use of electromyographic signal (EMG) that occurs during a muscle contraction. It is proven that four basic grasp shapes can be recognized from a two-channel EMG signal. The EMG signal is measured through the use of surface electrode. The EMG signal must be preprocessed, which means the filtration, amplification, and signal reconstruction. The EMG signal in original time domain contains huge amount of data and therefore it is not suitable for classification purposes. The EMG signal is processed in a digital signal processor (DSP) where the feature vector by the means of suitable transformation from signal in original time domain is obtained. For signal classification neural networks and artificial intelligence algorithms are used. The actual state of myoelectric prosthesis hand development, description of the electronic control system architecture, and summary of obtained results are described.

This article deals with development of a myoelectric prosthetic hand, that is developed at The Institute of Production Machines, Systems, and Robotics, FME, Brno University of Technology. A computer model of the prosthetic hand was proposed, first mechanical prototype made, and computer modelled mechanical characteristics verified on this mechanical prototype. The mechanical prototype was found applicable for serial production and was presented on The 47th International Engineering Fair at Brno, Czech Republic 2005. Nowadays development of a electronic control system for this prosthetic hand has been started. The prosthetic hand will be controlled through the use of electromyographic signal (EMG) that occurs during a muscle contraction. It is proven that four basic grasp shapes can be recognized from a two-channel EMG signal. The EMG signal is measured through the use of surface electrode. The EMG signal must be preprocessed, which means the filtration, amplification, and signal reconstruction. The EMG signal in original time domain contains huge amount of data and therefore it is not suitable for classification purposes. The EMG signal is processed in a digital signal processor (DSP) where the feature vector by the means of suitable transformation from signal in original time domain is obtained. For signal classification neural networks and artificial intelligence algorithms are used. The actual state of myoelectric prosthesis hand development, description of the electronic control system architecture, and summary of obtained results are described.

myoelectric signal, prosthetic hand, electromyography (EMG), neural network, artificial intelligence, EMG signal analysis

2006

14.06.2006

Orgis s.r.o

Praha: CTU

80-86059-44-8

Advanced Engineering Design 2006

Nezařazené články

1

6