Skip to content

Using HappyFeat in a BCI experiment

A BCI experiment consists of three steps:

  • signal acquisition using EEG hardware

  • signal analysis (realized using HappyFeat's main GUI)

  • online BCI with trained classifier, using EEG hardware

Steps 1&3 are also facilitated by HappyFeat. This section describes how to interface with EEG acquisition systems, using HappyFeat with OpenViBE.

Step 1: Acquisition

Setting up the experiment

  1. Make sure everything is ready for acquiring EEG data!
    • EEG cap set up
    • plugged to a computer with OpenViBE 3.5.0 installed
    • drivers set up in the OpenViBE acquisition server
    • etc.
  2. Launch the OpenViBE designer (using openvibe-designer.cmd (or .exe or .sh) in your install folder).
  3. Launch the OpenViBE acquisition server (using openvibe-acquisition-server.cmd (or .exe or .sh) in your install folder).
    • Connect and run the server
  4. Run happyfeat_welcome.py, and either start a new workspace or select an existing one.
  5. In your workspace folder, open the scenario file sc1-monitor-acq.xml.

Note: as explained in the Workspaces section, we strongly advise to create one workspace per experiment.

Note: the BCI experiment sequencing is set up in the mi-stimulations.lua script. Please refer to the tutorials on the OpenViBE website to learn how to fine tune your BCI experiment.

Data Acquisition

When you are ready, click "Run" in the OpenViBE designer. The BCI experiment is carried out, with instruction provided on the screen. EEG signals are written in the signals folder of your current workspace, with a filename indicating date and time of the acquisition.

Step 3: Online classification / Testing

  1. Make sure everything is ready for acquiring EEG data! (same as Acquisition step...)
  2. Make sure the OpenViBE acquisition server is running.
  3. In HappyFeat's main GUI, select a training attempt in the list in the lower-right part, and lcick on Use selected classifier. The scenario sc3-online.xml in the current workspace is automatically updated with the trained classifier weights, and set of features used for training.
  4. In the OpenViBE designer, open the scenario file sc3-online.xml located in your current workspace folder.
  5. Click "Run" in the OpenViBE designer. Instructions are provided to the user, along with visual feedback on the classification performance.
  6. EEG data is recorded in the signals folder for replay purposes.
  7. Classification results are recorded as CSV files in the results folder of the current session.

Note: the BCI experiment sequencing is set up in the mi-stimulations.lua script. Please refer to the tutorials on the OpenViBE website to learn how to fine tune your BCI experiment.