Muscle Synergy Analysis of Hand Motion Using sEMG

Overview

This project presents a framework for analyzing human hand motion using Surface Electromyography (sEMG), to capture and analyze muscular activity.

The extraction of muscular synergies happens through unsupervised learning methods, both linear such as NMF (Sparse and Classical) and PCA and nonlinear techniques such Autoencoders.

Applications include rehabilitation, prosthetics, human–robot interaction.

Project Structure - Code

• analyzer : folder containing scripts to better understand the main mathematical concepts (folded in classes) used such as Autoencoders, NMF (Classical and Sparse), PCA, Loader auxiliary functions, Error auxiliary function.
• config : file that contains all all the needed libraries and yaml configuration for project.
• datasets : folder containing the data used for the analysis (pinch, ulnar, power).
• helper : folder containing all the functions personally created for making the main scripts and analyzer classes easier to understand.
• tests : folder containing tests for synergy extrction based on all the different approaches presented.

Author

Riccardo Serraino

Internship at University of Bologna, Feb-Apr 2025

Supervisors: Prof. Roberto Meattini, Alessandra Bernardini