AbstractTraditional myoelectricity-based systems have inherent limitations and weaknesses to control dexterous prosthesis. The gap lies not in the methodologies that already extensively researched, but the means of signal extraction from the forearm. This thesis has explored and successfully tackled the practical control problems of dexterous prostheses using Sonomyography (SMG). Two hypotheses were made and supported by relevant literature to highlight the type of contraction that relevant to the amputees. The relevance of Fatigue-less Maximum Isotonic Contraction (FLMIC) phase is highlighted inSMG with experimental data. A novel wearable and portable SMG capturing system is presented with performances on a par with myoelectric methods. The Quasi-radial construction of the ultrasound transducer array allows reading A-mode signals from both anterior and posterior compartments of the forearm. The arrays comprised of transducers which are purpose designed to meet the requirements. The experiments with amputee and healthy subjects revealed that comparable gesture recognition accuracies can be achieved.
Proportional control of prosthesis is a noted problem. Since the majority of morpho-logical changes occur in isotonic or dynamic region, the tension produced by the muscleis low. Contrary to sEMG, where it requires a significant motor unit activation, a low level muscle activation is enough to provide proportional control using sonomyography.This phenomenon is investigated in this thesis. The cross-correlation method has been employed to recognize gestures between test and training sets.The experimental results demonstrated the ability to utilize the system in under water without significantly compromising the performance. The experiment also demonstrated that the effectiveness of Oil based coupling medium in such conditions. In this study a novel wearable ultrasound hardware is presented with evidence to prove its comparable performances with myoelectric systems.
|Date of Award||Sep 2015|
|Supervisor||Honghai Liu (Supervisor), Zhaojie Ju (Supervisor) & Steve Hand (Supervisor)|