Electrical and Computer Engineering Team14: BMW Brainwave Manipulated Wagon Midway Design Review.

31
Electrical and Computer Engineering Team14: BMW Brainwave Manipulated Wagon Midway Design Review

Transcript of Electrical and Computer Engineering Team14: BMW Brainwave Manipulated Wagon Midway Design Review.

Electrical and Computer Engineering

Team14: BMW

Brainwave Manipulated Wagon

Midway Design Review

2Electrical and Computer Engineering

Team 14 Members

Zijian ChenCSE

Tiffany JaoCSE

Man QinEE

Xueling ZhaoEE

Faculty Advisor: Qiangfei Xia

3Electrical and Computer Engineering

Outline

Review System Requirement

Block Diagram

Individual Responsibility

MDR Demo

CDR Schedule

4Electrical and Computer Engineering

Review: Problem Statement

People with physical disability is relatively dependent on others

•Limited strength to control wheelchair

Solution: BMWDemonstrates brainwave (EEG) control using robotic car

5Electrical and Computer Engineering

Outline

Review

System Requirement Block Diagram

Individual Responsibility

MDR Demo

CDR Schedule

6Electrical and Computer Engineering

What is BMW?

A brainwave controlled robotic car!

7Electrical and Computer Engineering

System Requirement

Computer application• GUI

• Database

Compatible with different user• training system

Remote Car Controls:• move forward, speed up, stop, and etc.

8Electrical and Computer Engineering

Outline

Review

System Requirement

Block Diagram Individual Responsibility

MDR Demo

CDR Schedule

9Electrical and Computer Engineering

Block Diagram

ComputerArduino/

C# Application

USB serial WriteMySQL

DatabaseXBEE:TX

Signal Processing

Command Algorithm

User Interface

XBEE: RX

ArduinoMotor

Power Supply

Neuroskyheadset

Bluetooth v3.0

ThinkGear Packet

Robotic Car

TX Arduino Module

Man Qin

Zijian Chen

Xueling Zhao

Tiffany Jao

10Electrical and Computer Engineering

Block Diagram

ComputerArduino/

C# Application

USB serial WriteMySQL

DatabaseXBEE:TX

Data Analyze

Command Algorithm

User Interface

XBEE: RX

ArduinoMotor

Power Supply

Bluetooth v3.0

ThinkGear Packet

Robotic Car

TX Arduino Module

Man Qin

Zijian Chen

Xueling Zhao

Tiffany Jao

Neuroskyheadset

11Electrical and Computer Engineering

Outline

Review

System Requirement

Block Diagram

Individual Responsibility

MDR Demo

CDR Schedule

12Electrical and Computer Engineering

Data Retrieval: Neurosky Headset

Neurosky Headset Laptop

Bluetooth v3.0

Thinkgear.dill

• Poor Signal

• Concentration/

Meditation

• Alpha/Beta/Theta

Power

• Raw Data

Retrieve attention level and corresponding raw EEG data

Save .txt file for further analysis

13Electrical and Computer Engineering

Graphical User Interface

Development language: C#

Warning message

Stimuli

14Electrical and Computer Engineering

Graphical User Interface: State Diagram

15Electrical and Computer Engineering

Command Algorithm

High Attention

List

Sorting

Calculate Classify Point

Attention Level

Low Attention

Concentrated Not Concentrated

Two consecutive low attention level

Input point > 90 % * classify point

16Electrical and Computer Engineering

Data Collection and Analysis

Why?

• Multiple Commands

• Achieve more reasonable algorithm

17Electrical and Computer Engineering

Data Collection and Analysis: Assumption

Different kinds of concentration:

• Increased memory – High Theta (3.5 – 6.7 Hz)

• Increased focus and awareness – High Alpha (11-14Hz)

• Increased conscious thinking – High Beta (14-30 Hz)

"Brain Wave States & How To Access Them.” Brainwaves Frequencies Change States of Consciousness. 1 Jan. 2005. Web.

22 Nov. 2014.

18Electrical and Computer Engineering

Data Analysis: Graph

*data is provided by the headset

19Electrical and Computer Engineering

Data Analysis: Challenges

No standard experiment is defined during data acquisition

Power spectrum is less stable than attention level

No way to know the time delay(FFT)

Power Spectrum are provided once per second (FFT)

20Electrical and Computer Engineering

FFT

Replace the attention level with other

parameters

Corresponding alpha wave and attention level

comparison

Matlab – perform FFT• C# will replace Matlab for FFT

Provide a platform for further experiment

21Electrical and Computer Engineering

FFT Result: Alpha Spectrum vs Attention Level

22Electrical and Computer Engineering

Outline

Review

System Requirement

Block Diagram

Individual Responsibility• Software Training System

• Research

MDR Demo CDR Schedule

23Electrical and Computer Engineering

Proposed MDR Deliverables

Demonstration of functioning algorithm• Controlling LED on/off • Stable result

Demonstration of graphical user interface• User determine control time• Display attention level

24Electrical and Computer Engineering

Outline

Review

System Requirement

Block Diagram

Individual Responsibility• Software Training System

• Research

MDR Demo

CDR Schedule

25Electrical and Computer Engineering

Proposed CDR Deliverables

Remotely control robotic car

Improved Command Algorithm

Utilize FFT result and classifier• Try to replace attention level

More user-friendly training interface• Graph• Database for training interface

26Electrical and Computer Engineering

Schedule: Ghatt Chart  11/24/201

411/30/201

412/7/201

412/14/201

412/21/201

412/28/201

41/4/2015

1/11/2015

1/18/2015

1/25/2015

2/1/2015 2/8/20152/15/201

52/22/201

5MDR draft report (All)                            MDR final report (All)                            XBEE TX/RX                            Robotic Car                            integration of robotic car and TX/RX                            

integration of RX and GUI (Tiffany+Man)                            

define experiment method (All)                            

FFT in c#                            (Beysian) Classifier                            Integration of FFT & Beysian with Training interface (Xueling+Tiffany+Zijian)

                           

Peforming Experiment with alrogithm (Xueling+Tiffany+Zijian)

                           

Database                            GUI improvement                            

27Electrical and Computer Engineering

Thank you

Any Question?

28Electrical and Computer Engineering

PDR Feedback

Why not use an embedded solution?

We decided to build C# application instead of embedded solution• Need to experiment to find a usable stimuli• Easy to set up an experiment interface with C#

application

29Electrical and Computer Engineering

Previous Ver.

30Electrical and Computer Engineering

Naïve Bayes Classifier

Probabilistic classifier based on applying Bayes’ theorem.

Bayes Theorem:

Advantage :• Better performances when the amount of input is large

• Easy to implement and requires O(n) run time.

• A way to determine alpha and beta relationship with attention level

31Electrical and Computer Engineering

How To Use Naïve Bayes Classifier

P (High Attention) = 40 /60 P (Low Attention) = 20 / 60

Now , we have a new input X

P (X | High Attention)=1/ 40

P (X | Low Attention) 3 /30