Optimizing the acquisition and processing of surface electromyographic signals

Electromyographic (EMG) signals, detected on the skin surface over active muscles, are being used increasingly for the control of prostheses and for investigating and retraining human movement. However, the statistical characteristics of the signal make it difficult toi achieve a processing delay of less than about 100 ms, which is marginal in some applications. Two techniques for reducing this delay are described: (i) signal pre-whitening, based upon digital autoregressive modelling or analogue filtering; an (ii) the use of a multiple electrode array. Successful discrimination between several states in a single muscle is possible in about 50 ms.