American Sign Group E Language Assistant Isabel Atanacio - EE - - PowerPoint PPT Presentation

american sign
SMART_READER_LITE
LIVE PREVIEW

American Sign Group E Language Assistant Isabel Atanacio - EE - - PowerPoint PPT Presentation

American Sign Group E Language Assistant Isabel Atanacio - EE Colin Fox - EE (ASLA) Charlie Munoz - EE Daniel Sarmiento - CpE Motivation Group Background: Allows for Extensive Software work as well as Hardware implementation.


slide-1
SLIDE 1

American Sign Language Assistant (ASLA)

Group E

Isabel Atanacio - EE Colin Fox - EE Charlie Munoz - EE Daniel Sarmiento - CpE

slide-2
SLIDE 2

Motivation

  • Group Background:

○ Allows for Extensive Software work as well as Hardware implementation.

  • Develop something that truly helps society and those with disabilities.
  • Desire to learn about:

○ Microcontrollers, Sensors, Wireless Communication, Data Management, Machine Learning.

  • Improve previous projects.
slide-3
SLIDE 3

Goals and Objectives

  • Implement 3 basic American Sign Language Parameters:

○ Hand Shape, Orientation, Location

  • Develop one Glove to translate hand gestures.

○ Letters

  • Translate from gestures to text and speech
  • Wireless Communication between gloves and computer
  • Open for addition of new gestures to library
slide-4
SLIDE 4

Project Specifications

  • Flexible → Easy hand gestures.
  • Weight = 2 lbs or less.
  • Small PCB within a 2”x2” square.
  • Bluetooth data transfer to interface.
  • Implement all 26 letters.
  • Text → Speech.
slide-5
SLIDE 5

Hardware

slide-6
SLIDE 6

Hardware Subsystems

  • Sensors
  • Wireless Communication
  • Microcontroller
  • Power
slide-7
SLIDE 7

Hardware Block Diagram

slide-8
SLIDE 8

Hardware Parts

slide-9
SLIDE 9

Sensors - Flex Sensors

  • 4.5’’ and 2.2’’ Flex sensors
  • Power rating : 0.50 W continuous
  • Linear Potentiometer
  • Life Cycle > 1 million Flexes
  • Manufacturer: Spectra Symbol
  • Cost: $7.95-$12.95
slide-10
SLIDE 10

Sensors - IMU

  • LSM6DS3: 6D orientation

■ Gyroscope and Accelerometer

  • Dimensions: 2.5 x 3 x 0.83 mm
  • Supply voltage: 1.71 V to 3.6 V
  • Operates at 3.3V 1.25mA
  • Manufacturer: ST Microelectronics
  • Cost: $3.54
slide-11
SLIDE 11

Sensors - Contact Sensors

  • Homemade Copper Tape Contact Sensors
  • Switch functionality
  • Functionality:

○ Helps Differentiate similar signs

slide-12
SLIDE 12

Wireless Communication

  • Bluetooth vs WiFi ?

○ Bluetooth uses 97% less power

  • Bluetooth

○ RN 41: 100m range, 2400-115200 bps

  • Voltage supply: 3.3V-6V

○ Power Consumption:

  • 25mA
  • Manufacturer: Roving Networks
  • Clost: $21.70
slide-13
SLIDE 13

Microcontroller - ATmega328P

Controls the flow of data from the sensors to the computer.

  • Dimensions: 7 x 7 x 1.2 mm
  • Operating Voltage: 1.8 - 5.5 V
  • Large Online Community
  • 7mA draw
  • Manufacturer: Atmel
  • Cost: $6.95
slide-14
SLIDE 14

Power - Battery

Lithium Polymer Battery

  • 1200mAh
  • Chosen based on

estimated power consumption

  • Estimated 4 hours of

runtime

slide-15
SLIDE 15

Power - Charger and Voltage Regulator

Powercell Board

  • Microchip MCP73831/2

○ Miniature Single-Cell Charge Controller ○ Control chip for battery charging ○ Price: $0.56

  • Texas Instruments TPS61200

○ Low Voltage Boost Converter ○ Voltage regulator ○ Operating Voltage : 0.3 - 5.5 V ■ 300 mA at 3.3 V, 600 mA at 5 V ○ Price: $4.49

slide-16
SLIDE 16

Power - LiPo Voltage Monitor

Maxim MAX17043G+U

  • Operating Voltage: 2.5 - 4.5 V
  • ±30 mV Accuracy up to 10 V
  • Typical Current Draw: 50 A
  • I2C Interface to MCU
  • Manufacturer: Maxim Integrated
  • Cost: $3.28
slide-17
SLIDE 17

MCU

slide-18
SLIDE 18

BLUETOOTH

slide-19
SLIDE 19

VOLTAGE REGULATOR

slide-20
SLIDE 20

BATTERY CHARGER

slide-21
SLIDE 21

LiPo VOLTAGE MONITOR

slide-22
SLIDE 22

IMU

slide-23
SLIDE 23

FLEX SENSORS AND CONTACT SENSORS

slide-24
SLIDE 24

Software

slide-25
SLIDE 25

Software Block Diagram

slide-26
SLIDE 26

Embedded Software Block Diagram

slide-27
SLIDE 27

Updating Glove Angles

  • Using Accelerometer and Gyroscope data

Θaccel=atan(accelZ/accelX) Θnew=α(Θold+Θgyro*dT)+(1-α)(Θaccel)

slide-28
SLIDE 28

Learning Block Diagram

slide-29
SLIDE 29

Translating Block Diagram

slide-30
SLIDE 30

Machine Learning

  • Support Vector Machines

○ Supervised classification learning ○ Very good with high dimensional data ○ Commonly used for gesture recognition

slide-31
SLIDE 31

Support Vector Machines

  • Separates different classes
  • f data with minimum error

margin

  • SVM are able to create

non-linear fittings

slide-32
SLIDE 32

Support Vector Machines

  • With more training sets the

fitting gets tighter and more irregular

slide-33
SLIDE 33

Python vs C++

C++ + Creates an executable + Faster at runtime

  • Memory management
  • More code needed for

implementation Python + Easy, straightforward syntax + No memory management + High-level native data types

  • Requires python and

libraries

  • Slower than C++
slide-34
SLIDE 34

Project Budget

Item Quantity Cost/Unit Total 6 DoF IMU 1 $3.54 $3.54 Battery Voltage Monitor 1 $3.28 $3.28 Flex 4.5” 4 $12.95 $51.80 Flex 2.2” 1 $7.95 $7.95 Bluetooth 1 $21.70 $21.70 Dev Board 1 $6.00 $6.00 LiPo Battery 1 $12.95 $0.00 PCB and parts 1 $157.37 $251.64

slide-35
SLIDE 35

Team Distribution

Sensors Bluetooth Microcontroller Embedded Software Computer Software Isabel P S Colin S P S Charlie P S Daniel P P

slide-36
SLIDE 36

Questions?