Control of a Bionic Hand based on Neural Networks and improved Gesture Recognition Techniques using multiple EMG sensors

This paper presents the development and implementation of a control algorithm, based on neural networks, intended for the control and operation of a small robotic prosthesis. The gesture recognition system is based on multiple surface EMG sensors. In this way, it has been possible to control the developed prototype in a simple and effective way. These EMG sensors serve as a bridge to detect small electrical pulses caused by muscle activity and transform them into an analogue signal that the control module is capable of filtering and interpreting. Final Year Project (TFG) or Final Master Project (TFM), constitutes an academic activity that allows us to evaluate in detail the different competencies acquired by the students during their time at university. In general, the development of robotic systems always supposes an extra motivation in the students. Thus, this document presents a novel and current electronic application such as the activation and control strategy of a bionic prosthesis. The purpose has been to simulate the actual movement of the fingers. At the same time, the different concepts and details of their implementation are discussed. The application of neural networks as a control strategy for the bionic hand has allowed the development of real-time control, obtaining very satisfactory results.