BioGUI

Modular PySide6 GUI for acquiring and visualizing bio-signals from different sources.

BioGUI Requirements

• uv (pip install uv or see installation instructions)

BioGUI Setup

To install dependencies, run:

uv sync

Development

To install the pre-commit hooks (ruff, black, prettier, license headers), run:

uv run pre-commit install

Run

Run the script main.py, which launches the main window.

You can run the application in two ways:

Option 1: Run with uv:

uv run main.py

Option 2: Activate the virtual environment and run directly:

source .venv/bin/activate
python main.py

Interface with board

To enable communication between the GUI and a board, one must provide a Python file with the following specifications:

• packetSize: integer representing the number of bytes to be read;
• startSeq: sequence of commands to start the board, expressed as a list of bytes;
• stopSeq: sequence of commands to stop the board, expressed as a list of bytes;
• sigInfo: dictionary containing, for each signal, a sub-dictionary with:
  • fs: sampling rate (float)
  • nCh: number of channels (int)
  • signal_type: dictionary with signal type information (required), must contain at least:
    • type: signal type, either "ultrasound" or "time-series" (string)
• decodeFn: function that decodes each packet of bytes read from the board into the specified signals.

Some examples of interface files are provided in the interfaces folder.

Utilities

In the utils folder, there are some utility scripts: the most useful one is plot_signal.py, which shows how to open the .bio binary file containing the acquired signals.

Applications

This repository contains additional components in the applications/ folder:

• bio-bridge — BioBridge: real-time ML inference middleware
• motion-lab — MotionLab: Unity environment for hand control and task evaluation

To run the full gesture-control pipeline, additional setup is required for BioBridge and MotionLab. See bio-bridge/README.md and motion-lab/README.md.

Preparation (order does not matter):

• Open BioGUI and configure the interface and forwarding settings.
• Open MotionLab in Unity and load a scene (do not press Play yet).

Start (in this order):

1. Run BioBridge — waits for an incoming BioGUI connection.
2. Start acquisition in BioGUI with forwarding enabled — data is displayed and forwarded to BioBridge.
3. Press Play in Unity — receives hand pose data from BioBridge and starts rendering.

Live Demonstration

tasks.mp4

Authors

This work was realized mainly at the Energy-Efficient Embedded Systems Laboratory (EEES Lab) of University of Bologna (Italy), and at the Digital Circuits and Systems (IIS) of ETH Zurich by:

• Mattia Orlandi (University of Bologna)
• Pierangelo Maria Rapa (University of Bologna)
• Enzo Baraldi (ETH Zurich)

Citation

If you would like to reference the project, please cite the following paper:

@ARTICLE{10552147,
  author={Orlandi, Mattia and Rapa, Pierangelo Maria and Zanghieri, Marcello and Frey, Sebastian and Kartsch, Victor and Benini, Luca and Benatti, Simone},
  journal={IEEE Transactions on Biomedical Circuits and Systems},
  title={Real-Time Motor Unit Tracking From sEMG Signals With Adaptive ICA on a Parallel Ultra-Low Power Processor},
  year={2024},
  volume={18},
  number={4},
  pages={771-782},
  keywords={Electrodes;Real-time systems;Muscles;Motors;Electromyography;Circuits and systems;Graphical user interfaces;Blind source separation;human-machine interfaces;independent component analysis;low-power;machine learning;on-device learning;online learning;PULP;surface EMG},
  doi={10.1109/TBCAS.2024.3410840}}

License

All files are released under the Apache-2.0 license (see LICENSE).