By:
Fatemeh Akrami
Supervisor:
Dr. Hamid D. Taghirad
October 2017
Brain Computer Interface Control of a Virtual Robotic
System based on SSVEP and EEG Signal
Contents
Introduction
Material and Methods
System Integration
Experiment results
Conclusion
• A direct connection pathway between the brain and external world.
• BCI acquires brain signal in response to a certain type of behavior and
decode it into the device control commands.
Introduction
Material and Methods
System Integration Experiment
results Conclusion
Brain Computer Interface (BCI)
Brain Computer Interface (BCI)
• Response to flickering stimulus.
• Generates in the visual cortex of the brain.
• It contains stimulus frequency and its harmonics.
Introduction
Material and Methods
System Integration Experiment
results
Steady State Visual Evoked Potential
Brain Computer Interface (BCI)
Introduction
Material and Methods
System Integration Experiment
results Conclusion
PSD of SSVEP Signal
Amplitude spectrum of the SSVEP BCI signal for P7, O1, O2, P8 electrodes in 17Hz stimulus frequency
Brain Computer Interface (BCI)
• SSVEP needs less training time
• Information transfer rate is high
• It has higher classification accuracy
Introduction
Material and Methods
System Integration Experiment
results
Why SSVEP signal?
Brain Computer Interface (BCI)
Introduction
Material and Methods
System Integration Experiment
results Conclusion
SSVEP based BCI system
Typical structure of the SSVEP based BCI system
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Visual stimulus
Introduction
Visual Stimulus Design
Mechanical design MCU and control color and frequency Connection between
MCU and PC
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results Conclusion
Visual stimulus
Introduction
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Data acquisition
Introduction •
14 channels: AF3, F7, F3, FC5, T7, P7, O1, O2, P8, T8, FC6, F4, F8, AF4
• References: In the CMS/DRL noise cancellation
• Sampling rate: 128 SPS
• Resolution: 14 bits 1 LSB = 0.51μV
• Bandwidth: 0.2 – 43Hz
• Wireless and rechargeable
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results Conclusion
Data acquisition
Introduction
• Distance between the stimulus and subject is 80 [cm].
• Visual stimulus frequencies are 11, 13, 15 and 17 [Hz].
• The color of the stimulus is set to green.
• Electrodes placed on O1, O2, P7 and P8 locations.
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Data Processing
Introduction
M = mean value
C = Covariance matrix
The reference signal
Likelihood Ratio Test (LRT) method
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results Conclusion
Data Processing
Introduction
Likelihood ratio for static signal
The measure of association
L=1: data are perfectly correlated L=0: data are uncorrelated
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Virtual robotic arm
Introduction
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results Conclusion
Virtual robotic arm
Introduction
RV2AJ Simulink model
• The robot has two degree of freedom.
• Theta1 and Theta2 are inputs to determine the robot movement.
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
System Integration
Introduction
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results Conclusion
State of the robot
Introduction
State of the robot corresponding to stimulus frequency
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Offline results
Introduction
• Data was recorded in ARAS lab
• Four subject participated in experiments
• Length of the data is 20 second in four different trails
• The experiment was performed in dim room
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Conclusion
Offline results
Introduction
• Each time window for processing is four second
• Accuracy is the number of the correct segments to the total
number of segments
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Online results
Introduction
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Conclusion
Online results
Introduction
Online result of LRT algorithm and Classification output in test2
Brain Computer Interface (BCI)
Material and Methods
System Integration Experiment
results
Conclusion
Introduction • An effective strategy is done for implementation of complete BCI system.
• LRT method is evaluated in offline analysis and implemented for online detection.
• It is improved that the online detection with high accuracy in short time windows are possible.
• It is improved that the input stimulus frequencies are perfectly distinguishable.
• The experiments show promising features of the developed system for further application.