| This thesis investigates the extraction of information from a single bipolar myoelectric (ME) signal. Currently, ME signals are used to control artificial limb prostheses clinically. Very little processing of the ME signal, however, is used in conventional systems. The aim of this thesis is to investigate various sophisticated ME signal processing strategies to develop a clinically viable alternative to the conventional ME controlled limb prosthesis. The strategies investigated combined neural-net and spectral processing. These approaches were compared in a preliminary investigation to determine which is the most promising ME signal processing approach for implementation. Implementation of this approach in real-time then allowed clinical measures to be used in its analysis. Comparison of the results of this analysis with clinical results of conventional systems indicated the viability of the more sophisticated scheme for use in a Prosthetic Controller. (Abstract shortened by UMI.). |
