Skip to content
Charles Bélanger Nzakimuena edited this page Apr 14, 2019 · 4 revisions

Welcome to the MindPong wiki!

Introduction

Our student club’s flagship demonstration brain-computer interface, MindPong, offers the means for players’ brain waves to communicate with a computer and directly see the impact of quick changes in brain activity. Unlike a pioneering mind-controlled game of the same name before it which drew inspiration from the well-known arcade video game ‘Pong’ (https://vimeo.com/105356035) , our MindPong was created after the actual game of ‘Ping Pong’. The goal of the two-player game is to physically move a real ping pong ball in your opponent’s direction using none other than your own brain-waves.

MindPong Overview

Since its creation by some of the founders and very first members of PolyCortex, MindPong has been continuously improved upon and optimised. As an EEG signal-driven multiplayer game, it requires two muse headbands and an Arduino UNO microcontroller board. Upon a wooden support rests a plastic tube at the end of which two fans are located. An opening at the midpoint of the tube allows a ball to be introduced inside the tube. The EEG signal from each muse headband is communicated to a computer via bluetooth. The computer software is linked to the microcontroller which carries the signal that activates the fans based on EEG activity. The goal is for one player to move the ball towards one end and the other player to move the ball to the other end. MindPong is designed to allow the selection of specific types of EEG activity and once selected, the greater the amplitude of the signal of a given muse headband, the more its corresponding fan rotates. The game relies entirely on the pyMuse library, maintained by PolyCortex and available on our github page (https://github.com/PolyCortex/pyMuse).

Muse 2016

Muse is a direct-to-consumer EEG headset and the device’s performance for use in research settings has been validated [1]. Muse electrode locations follow the 10-20 international EEG placement system. The muse 2016 itself provides a sampling rate of 256 and Muse Direct (or MuseIO) allow the user to track the activity of individual EEG activity from the headbands.

Interface

The MindPong interface was completely redesigned in the past year. The main tab shows the start button to initiate the game. An arithmetic-based mathematical problem is presented to the players with the intend of initiating a significant change in their brain activity. Two modes are available, first, the drill exercise which provide simpler operations. Also available is the fill in the blank which is more challenging. Each category has a difficulty drop down menu to make them more accessible to players of all ages and mathematical abilities. The relaxation mode instructs the players to simply relax. This mode is primarily driven by the expectation that higher frequency activity will progressively reduce as the players relax, and the player with the lowest activity wins the game. Once the Start button is pressed, an arrow shows the direction of the ball and reflects the player with the dominant selected brain activity.

dassads

Figure 1 MindPong interface main tab

The analysis tab features static data time domain and spectrogram visualisation. The data allows the operator and players to determine if a significant response was observed corresponding to answering mathematical problems. It allows our team to continuously improve upon its EEG activity detection algorithm, to make it more responsive to normal human responses to mathematical activity with anonymized data. The game date and time are also recorded. The settings tab allows the basic configuration of the selected muse UDP port, and the serial communication port for interfacing with the Arduino microcontroller.

fafsfs (Custom)

Figure 2 MindPong interface analysis tab

Electrical circuits

We use two independent circuits to make the game work. The first one allows the arduino to control the fans and the second one allows the microcontroller to raise a cute little flag for the winner.

Fan control circuit

The circuit makes it possible to use a PWM to control the fans' speed. To do so, a transistor (N-MOSFET) blocks and enables the 12V current to flow to the fans. We are using a flyback diode to provide a path to dissipate energy stored by the motor inductance. Towards continuing to improve MindPong’s design and reproducibility, this year the fan prototype circuit was turned into a PCB. The schematic and layout of the PCB are open source and are available on our repository. (https://github.com/PolyCortex/MindPong).

Figure 2

Figure 3 Fans PCB DipTrace circuit schematic and layout

fansCircuit (Custom)

Figure 4 Assembled fans PCB circuit

Lasers and laser receivers

Also added to MindPong in the past year is a ball laser detection system. When the ball crosses the laser detection area at the extremity of the tube on the opposite side of the player, the flag is raised to indicate the player’s victory. The flags circuit is composed of three components for each of the two flags: a laser emitter, a laser receiver and a servomotor with the flag affixed to it. They are all linked to the arduino's GPIOs (digital pins) and powered by the 5V output of the arduino.

Figure 3

Figure 5 Lasers and flags example setup diagram

Experimental Features

Auxiliary electrode

As part of our MindPong optimization efforts, our team set out to add to an electrode to each of our muse headsets. All the steps to adding an auxiliary electrode to muse EEG headsets were made available online in a tutorial originating from a member of the Los Angeles NeuroTechX chapter (https://hackaday.io/project/162169-muse-eeg-headset-making-extra-electrode/details). An additional available channel for each headset greatly expands the possibility of single derivation observations. The addition will also be useful when seeking to reproduce results or observe events that are highly localized spatially.

AuxTest2

Figure 6 PolyCortex members testing the auxiliary electrode

auxTest

Figure 7 PolyCortex members viewing the auxiliary electrode signal with the muse-lsl viewer

Beta mode

For example, wide reductions in alpha and beta frequency range activities, corresponding to 8 Hz and above, have been associated with math problem solving [2]. For our purposes, the added electrode should allow us to determine, over repeated trials, if a significant amplitude suppression of activity that surpasses 8 Hz can be observed in posterior locations when mental arithmetic is performed.

Figure 4

Figure 8 Occipital lobe and muse auxiliary electrode available placements within the 10-20 system

A clearer indication that mental change has occurred in response to thinking is important to us, as it allows players to more directly and volitionally influence their brain activity and therefore have control over the movement of the ball. To test the new electrode, we proceeded to perform the eye closure test with the auxiliary electrode positioned at a posterior location (Fpz-O2). As can be seen in 2D and 3D spectrograms, activity was raised in approximately the 10-12 Hz frequency range with sustained eye closure compared with the eyes open condition.

Figure 5

Figure 9 Eyes open and closed 2D and 3D spectrograms

While reproducing the electrode addition project was successful, the MindPong software does not currently receive data from the auxiliary electrode and we intend to integrate the feature in future iterations of the game. Higher than 8 Hz can still be selected as the game-driving signal using the headset’s standard electrodes.

Sleep mode and clinical applications

MindPong is more than just an entertaining and educational tool. Indeed, there are concrete applications to signal features extraction and classification with many carrying clinical implications. One such application, is the study of sleep. Our team began to develop tools towards leveraging MindPong technology and applying it to real world issues. A repository was created and features sleep database and Muse static data analysis and dataset generation (https://github.com/PolyCortex/SleepEEG).

References

  1. Krigolson, O., et al., Choosing MUSE: validation of a low-cost, portable EEG system for ERP research. Front. Neurosci. Brain Imaging Methods, 2017. 11(109): p. 1-10.
  2. Lin, C.-L., et al. Brain dynamics of mathematical problem solving. in 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 2012. IEEE.