A control unit for myoelectrically controlled prosthesis

The long time for getting an acceptable response in myoelectric prosthesis has been a problem. Thanks to DSP processors, the myoelectric signal (MES) processing time has had a drastic reduction. This paper describes a control unit for a myoelectric prosthesis using statistic variables related to contraction force. ANOVA analysis was used to seek the best separable characteristic set. A MES stationarity detector was implemented and a perceptron artificial neural network was used as classifier. A dedicated system was built using a DSP processor and it has a fast response.