Mind Games. Daniel Warner (EE) John Parker (EE) Justin Dwyer (EE) Duy Nguyen (EE) G38

Transcription

1 Mind Games Daniel Warner (EE) John Parker (EE) Justin Dwyer (EE) Duy Nguyen (EE) G38

2 Goals & Objectives Basic Overview The goal of this project is to control multiple applications through nontraditional methods. Human Input - Brainwaves, Muscle Movement, Heart Rate via BCI (Brain Control Interface) DSP - Laptop Computer to MCU Application - RC Car, Water Fountain, Claw Machine

3 Motivation Diverse Multi-discipline project - Aspects include: Biomedical Engineering DSP / Programing RF / Wireless Communications Mechanical Engineering Concepts

4 Specifications -Overall System Latency of 750ms -General controller board should be flexible and offer many control options -Claw machine needs to capable of carrying a load of 1/2 pound -Claw machine needs a minimum of half of an inch precision in x-y motion -Water fountain displays pulse across three pumps -RC Car should be drivable using basic controls (Left, Right, Forward, Reverse)

5 Overall Block Diagram

6 BCI Hardware Key Points -Multiple BioPotential Inputs -High Resolution -High Bit Depth

7 Electrodes Passive vs Active -Passive requires Capacitive gel to overcome trace impedance of skull -Active uses small amplifier to boost signal -Modern Active and Passive Electrodes have a similar SNR

8 Headwear Headset International Standard for EEG Capture Primarily Targeting Brain Alpha Waves Found between 9-14 Hz Strongest Propagation in the O1 and O2 Region Occurs when users eyes are closed

9 DSP Software 3 options for software choice -MATLAB, Python, BrainBay Pro Con MATLAB Extremely powerful, experienced in using it Expensive to buy software & Processing Package Python With correct modules, as powerful as MATLAB, free, Code establishing connection to board available No experience in programming python BrainBay Extremely simple to use, free Too simple. Lacks programming options & output is limited to keyboard strikes.

10 Lab Streaming Layer The lab streaming layer (LSL) is a system for collecting data. It handles both the networking, time-synchronization, access as well as optionally the centralized collection, viewing and disk recording of the data. We will use LSL to send our data from the board into the scripts designed to process it. Latency is a big concern with our project. LSL samples are transferred with less than.1 milliseconds of latency for samples smaller than a megabyte.

11 EMG The code for this portion must detect a muscle being flexed. The spikes created by this muscle activity are huge and easily detected. We will compute the absolute value of the signal, then output a positive if the signal is above 12uV.

12 EMG Output We will be monitoring the muscle activity of the jaw in the Claw Machine program. As soon as we detect a clenched jaw, we will output a C over serial to the PCB controlling the Claw Machine.

13

14 EKG Detection The code for this portion must detect the pulse of the user. By counting the number of samples between the strong negative peaks, the time between each pulse can be determined. Sample of minimum Samples between mins Calculated pulse

15 EKG Output Limitations on communicating between python and our MCU limited us to transferring ascii characters. Once the pulse is found, that number is converted into an ascii character, and that character transferred to the MCU

16 EKG Program Flowchart

17 Alpha Wave Detection Alpha waves are produced at ~10 Hz. The image below shows EEG Data band pass filtered between 9-12 Hz to show the alpha band. When the eyes are closed, there is a large increase in activity in the range 9-12 Hz.

18 Alpha Wave Detection Detecting Alpha -Filter Signal -Compute Average -Output if wave detected

19

20

21 Determining Head Tilt Determining the way a user is tilting their head will be accomplished using the on board accelerometers. Compute what percentage of earth s gravity each axis experiences From that, the angle of the head will be determined

22 Accelerometer Processing

23 Accelerometer Testing Data is shown in its separate channels, with channel 0 = x, 1 = y, 2 = z The data begins with the board flat on a desk, then tilted different directions every 5 seconds

24 Head Tilt Output If a head tilt is determined using the accelerometer, a coresponding ascii character is output. The Claw Machine will expect an l for left, r for right, f for forward, and b for backwards. The RC Car will expect a w for forward, an s for backward, a d for right, and an a for left

25 General Development Board Many specialized Dev boards are currently available for different technologies. 3G GSM, Bluetooth, Wifi, RF. We wanted something that was able to do more and offered useful inputs with a seamless connection.

26

27

28

29

30

31

32

33

34

35

36 Water Fountain;/,.,. This water fountain will change output based on the user s pulse. The following slide will show the schematic for the water fountain. Tiers active Pulse<75 75<Pulse<110 Pulse> ,2 1,2,3

37

38 Issues *Inductive load from relay end *sparking from contact end *inductive noise when pumps shut off *isolating analog and digital grounds *decoupling capacitors *solid state relays *dc motor

39 Claw Machine Application References not many claw machine references -decided to reference the x-y motion/rails from a 3d Printer -attach a motor on a platform for z motion User friendly: Easy input signals to control the claw machine game. with little to no BCI training. EMG signals: Unlocks control of x and y-axis Accelerometer signals: tilt forward -> move forward tilt backwards -> move backwards tilt left -> move left tilt right -> move right EEG signal: close eyes and relax (alpha wave) -> claw drop

40 Claw Machine Subsystem

41 Bipolar Stepper Motor Advantages: Provides high resolution and fair amount of torque, relatively cheap, long lifespan Disadvantages: Current draw, Dedicated Drivers, No internal feedback system Requirements: Project requires precision, enough torque to move the metal rails, and durability. Power consumption is not a main concern and a feedback system can be implemented through hardware and software design How to overcome the feedback issue: z axis:utilize a counter in software and rely on precision of stepper motor x-y axis: Timing belt slipping means counter is not an option Utilize limiting switches to send signal to mcu when collision occurs

42 Stepper Motors Leadshine s stepper motor selection document used for torque calculations through dynamics free body diagrams and equations: Stepper motors selected:

43 x-y Axis Feedback System Encoder: expensive hardware and software unnecessarily harder bulkier Limit switches: Cheap Simple configuration effective easy coding

44 Stepper Motor Driver Allegro A4988: Max load supply voltage: 35V Max output current: 2A Logic supply voltage: 3.3V to 5.5V Enable pin will be used to shut off holding torque and reduce power consumption Heatsink required for full 2 A current draw

45 Claw Relay Parasonic 5V Relay: JS1A-B-5V-F High contact capacity: 10 A Max operating voltage: 100V DC Nominal operating current at 5V: 72 ma Coil resistance: 69.4Ω FairChild NPN Transistor: BC337 Max Ic current: 800mA minimum hfe at Ic=100mA: 100 Vbe(sat): 1.2V Vce(sat) from circuit configuration: 40mV *ensure Ib does not exceed mcu rating

46 Power System RAIDMAX XT series RX-300XT 300W ATX12V / EPS12V Power Supply: V@14A, +5V@15A, +12V@21A cost: $22 Need: output 24V 1 Amp Current *attempted to use +12V to -12V for 24 but could only supply.5a *used 24V wall TI TPS55340RTER DC-DC Boost Converter: Vin: 2.9V to 32V Vout max: 38V 95% efficiency cost: $5

47 Program Flow x-y motion

48 Delivery System Program Flow

49 Work Distribution Daniel John Justin Bob BCI Hardware DSP Primary Secondary Secondary Primary Wireless Communication RC Car Water Fountain Claw Machine Admin/ Management Primary Primary Primary Primary Primary Primary

50 Budget

51 Financing / Sponsors

52