| Summary: | Electromyography, otherwise known as EMG, is the study of the electrical properties of our muscles. Significant amounts of funding and research have developed it into one of the leading technologies used in prosthesis, but it also shows tremendous potential as an EMG based human machine interface. However, it encompasses restrictions that prevent the general public from adopting the technology. This thesis focuses on improving EMG based Human Machine Interaction (HMI), in order to take a step towards making EMG systems easier to use for the general public. The nature of this research investigates novel approaches at reducing problems with multi-user interaction. Firstly, by investigating optimum electrode configurations with simple signal processing and machine learning to test an EMG gesture recognition system. Secondly, by using forearm circumference as an estimator of Maximum Voluntary Contraction (MCV), for use in an automatic threshold calibration technique, which includes the development of automatic circumference measuring armbands to make measurements and calibration completely autonomous. The third stage of this thesis investigates how EMG, circumference, and a combination of both, can be used for biometrically identifying a user. This benefits systems that require storing and retrieving past data or settings, and has promise for automatic retrieval of past training data. This is particularly useful for enhancing multi-user interactions, as each user will then be able to use their own previous data on an EMG system, which has been shown to be better than using other user's data. The final stage of this thesis creates an EMG and motion sensor fusion armband, to investigate how sensor fusion can enhance EMG interaction, with potential applications in robot tele-operation, Human Computer Interfaces (HCI) for amputees, and 3d object manipulation.
|
|---|
