Design of an underactuated hand prosthesisdevelopment platform and artificialneural network-based control

In this thesis, kinematic and dynamic analysis of a selected tendon-based underactuated hand prosthesis mechanism was performed. As a result of the analysis, a dynamic model was created by considering the interaction of the mechanical structure with an object in the workspace. Simulation outputs and a vision-based motion analysis platform were used to determine the performance of the dynamic model to model the dynamic behavior of the mechanical structure. The simulation and experimental results show that the dynamic model can successfully represent the dynamic behavior of the underactuated mechanical structure. In addition to kinematic and dynamic modeling of the hand prosthesis, the methods for the control of the prosthesis based on EMG signals were examined. Firstly, EMG signal features that can be used in hand motion classification based on EMG signals have been determined by the studies on an externally available dataset. Two different classification methods, in which these EMG signal features are used as input have been developed and the performance levels of the developed classification methods were determined by experimental studies. As a result of the experiments, it was found that hand movements assigned to 3 different EMG inputs can be successfully classified and realized.