This master's thesis focuses on classifying human gestures and finger postures using surface electromyography (sEMG) signals from the forearm. sEMG can non-invasively measure muscle activation and is widely used in prosthetic control and human-computer interaction. This project will place multiple wet electrodes around the forearm to collect muscle activity associated with different hand and finger postures.
The goal is to design a machine learning workflow to process sEMG signals and classify them into a set of predefined gestures or finger postures (e.g., clenching a fist, pinching, pointing, opening the hand, flexion and extension of a single finger). Tasks include signal preprocessing, feature extraction, classifier training, and performance evaluation.
Reliable gesture classification is fundamental to achieving intuitive prosthetic control and interactive robotic systems. This thesis delves into the relationship between forearm muscle patterns and hand postures, thereby enabling more natural human-computer interaction.