My professor told me to collect data for the states we want to to use to control the movement of our robot i.e., thinking, meditation and other emotional or physical states using an Emotiv EEG headset which I don't have at the moment.

Is there any alternative software for acquiring and visualizing these signals without the need for headset right now? How many samples need to be taken for the training of the neural set or any other classifier? What sampling rate and which file format must these samples need to be saved?

Is there any online free simulator for analyzing the data and map it’s working on a simulated robotic object? Which embedded hardware is ideal for the project Arduino or Raspberry Pi for a 32 bit Windows 7 operating system.

What online tutorials, research papers and books would assist in taking this project to completion and creating artificial neural networks?

  • 3
    $\begingroup$ There seems to be 6 questions in this one post. Would you mind breaking these into separate posts? Additionally, I don't understand what data you think you can acquire without a headset? Do you have an array of electrodes lying around? Are you talking about acquiring data from different modalities? $\endgroup$
    – Seanny123
    Jun 19, 2018 at 15:25
  • $\begingroup$ Look into it more. Depending on the Arduino, it can output signals (digital, 0-5V analog, serial)--but you need to know what signals the electrode or whatever type of sensors the headset uses. It sounds like you don't know what signal is used--so that's step one. As written, it seems like you haven't looked into it much. $\endgroup$
    – adamaero
    Sep 20, 2018 at 21:59
  • $\begingroup$ I don't think this is an appropriate question for this SE. Perhaps try the Electrical Engineering SE once you've done more research. Then, maybe even dsp.stackexchange.com $\endgroup$
    – adamaero
    Sep 20, 2018 at 22:03

1 Answer 1


There is lot of way to create BCI, (SSVEP, VEP, imaginary movement, brain state, etc...). In your case it seems to be with brain states.

You should first know what kind of information you will extract from your signal. In this case you will need to extract different band of frequencies (alpha, beta, gamma, theta, etc...). To do that, you can use calculate spectrum based on Fourrier transform, or use wavelet...

Then you will need to classify this different states, using clustering or machine learning techniques (ANN, naives bayes, PCA, SVM, etc...). You should have a 'training' dataset, where you have extracted information with state label.

Then you create a 'mapping' where for each detected state, and you get an activation of your effector (robot)

I recommend you to use python, where you have lot of library available for signal processing, and machine learning.

Good luck !

  • 1
    $\begingroup$ Try to source your answer by citing papers or linking to credible web sites. I do like the way you tackle a (way too) broad question, so +1, but try to back it up with links (e.g. to sites explaining the basics of EEG, waveletes, FFT etc). $\endgroup$
    – AliceD
    Jun 21, 2018 at 8:21

Not the answer you're looking for? Browse other questions tagged or ask your own question.