Gary O’ Donoghue Electronic & Computer Engineering, National University of Ireland, Galway Final Year Project A small number of consumer electronics EEG headset are available on the market primarily aimed at allowing gamers to interact with games by “thinking”. This project will focus on a more sophisticated task which involves the control of a model vehicle by means of the analysis of the EEG waveforms being recorded by a wireless EEG headset. The control of the vehicle will depend on the level of relaxation of the user, which can be determined by evaluating the Alpha wave power in the user’s EEG. This type of device can be used for people diagnosed with ADHD and ADD to improve focus and concentration. The Emotiv Epoc (shown right) reads a user’s EEG data using the 14 electrodes and transmits these values to the PC over wireless link. Each electrode records 128 EEG data samples every second. These samples are time domain voltage values representing fluctuations in the voltage at specific areas around the user’s scalp, based on the international system (shown right). Only two occipital channels’ data (O1, O2) are used to determine the state of the user’s eyes (open/closed). When the samples are read in to the PC using the headset, they may contain artifacts. Artifacts, highlighted in red in the figure below, are jumps or dips in the voltage values due to user eye movement and blinking. These results are undesirable and can be removed by a simple process of getting the averages of the EEG data every second (128 samples) and deciding whether this value is within or outside an acceptable voltage region, the latter resulting in the samples being discarded as artifacts. Once the artifacts have been removed from the samples, they then need to undergo frequency analysis, specifically Fourier Transform. This converts the time-domain voltage values from the headset to frequency domain values representing the energy or power at each frequency. These values can then be used to determine the power at specific frequencies e.g. the Alpha frequency band power can be obtained by averaging the frequency magnitudes between 8 and 12Hz. The figure below, which illustrates the variation of Alpha wave power over time, shows a jump when the user closes their eyes at the 30 second mark. PC running our software is connected to toy car via Bluetooth. Each second, data is read in from the wireless headset from electrodes O1 and O2, located near the back of the scalp. These values then undergo artifact removal and frequency analysis. Depending on the results of this frequency analysis, certain commands are sent over Bluetooth to control the movement of the toy car. Once this command has been sent to the car, the program starts reading in EEG data again. Emotiv EPOC headset Emotiv electrode locations with O1 and O2 highlighted The Bluetooth car consists of an Arduino microprocessor with a Bluetooth and Ardumoto shield connected to it. The car should only move when the person’s EEG indicates that they are in a relaxed state (when their Alpha power is above a certain threshold). Only when the PC software detects this, should it send a command to the car to start moving. The values from the frequency analysis are received by the Bluetooth shield and interpreted by the Arduino. The Arduino then uses the Ardumoto shield to power the motors, turning them on and driving the car if certain value is received. Plot of voltage with respect to time, with artifacts highlighted in red Figure showing the set up of the toy car with a schematic of the Arduino connected to the Bluetooth and Ardumoto shields Plot of Alpha wave power with respect to time Flowchart of the overall operation of the system