This thesis focuses on using surface electromyography (sEMG) signals to track and predict the complex movements of the human hand and arm for intuitive robotics interaction. Surface EMG measures the electrical activity of muscles through electrodes placed on the skin, providing direct insights into muscle activation patterns during motion. In this project, six wet electrodes will be attached across the forearm and upper arm to capture muscle activity related to both limb and finger movements. The main goal is to predict continuous motion trajectories of the entire arm and hand, instead of classifying discrete gestures in many previous studies. This requires developing signal processing and deep-learning methods capable of delineating the fine-grained, coordinated motions of both the arm and the hand in real-time. The predicted motion trajectory will be used to control a robotic arm and robotic hand, enabling dexterous interaction tasks. This research contributes to the development of more natural and precise muscle-driven robotic control systems for prosthetics, rehabilitation, and human–robot interaction.
