Project P12016

Magy Yasin |Dave Taubman |

Curt Beard |Oliver Wing |

Aalyia Shaukat |Stu Burgess |

Jeff Chiappone |

Multi-Disciplinary Senior Design 1

  Project Manager  Lead CE/EE  Project Engineer  Project Engineer  Project Engineer  Project Engineer  Lead ME/IE

Systems-Level Design Review | Oct. 7, 2011

Agenda• Meeting Goals • Project Background

o Objectiveso Customer Needso Spec Overviewo Team Overview

 • Functional Analysis

 • Concept Development

 • Physical System Architecture

o Arriving at this System • Risk Assessment

 • Schedule 

Goals

• Gain Feedback:  positive or negative specific feedback on design

 • Identify Problem Areas: based on feedback, develop a

road map of problem areas and how to solve them • Discover Ways to Improve:  our models, our

assumptions, our methods

• Answer the Question:  does our design do what our customer needs? 

 • Acknowledge Readiness:  to move onto detail design

10/6/11

"Design and Fabricate a Navigation Aid for the Visually Impaired/Blind"

 • Navigates a person from any point on the second floor of the

Gleason Building to any other point on the floor• Map is based on RFID tags which define rooms and landmarks• Utilizes Dijkstra's Routing Algorithm to determine optimal route• Device does not rely on auditory cues• Does not impair use of cane or guide dog

Project Objective

Customer Needs

Customer Need #

Importance Description

CN1 1 Navigates blind person from any location to any other location on a single floor within a building, where destination has a room number.

CN2 1 Navigates efficiently

CN3 1Operates successfully under conditions that the user is likely to encounter, such as: destination change while en-route, user goes "off the map," user stops using the device while en-route, device fails, ect.

CN4 1 Portable and untethered

CN5 1 Gives non-visual, non-Braille instructions/feedback

CN6 1 Easy to use

CN7 2 Easy to train user

CN8 1 Comfortable and safe to use

CN9 1 Operates for an entire day without recharging

CN10 1 Battery is easy to replace

CN11 2 Doesn't distract the user or others near the device

CN12 2 Doesn’t draw attention to the user

CN13 1 Hands are not needed to carry device

CN14 1 Parts cost < 700 dollars

CN15 2 Navigate blind person to drinking fountain, restroom, or other Important location that may not have a (known) room number

CN16 3 Work independently of system of units (SI vs English)

Device Specifications

Specification Unit of Measurement Ideal Value

Size Inches <= 4x2x1

Weight Pounds <= 0.025

System Cost Dollars < 700

No. of 20-min Navigation Intervals per Charge # >= 10

Charge Time Hours <= 8

Battery Replacement Time Minutes < 1

Map Read/Load Time Minutes < 1

Wrong Command Provided to User Frequency < 1/1000

Commands are Non-Visual Y/N Yes

Distinct and Distinguishable Commands # >= 5

Distinct Inputs From User # >= 4

Verification of User Input Provided Y/N Yes

Size of user input Digits 6

Duration Between Commands Seconds <=2

Device Specifications

Specification Unit of Measurement Ideal Value

Operating Conditions: Temperature Degrees (Celsius) 0-40

Operating Conditions: Relative Humidity Percentage 0-100

Noise Generated By Device at 3 ft Decibels < 50Impact Resistance: Fully Functional After Drop From

Set Height Feet 3

Hands Required to Carry System # 0

Hands Required to Position/Use System # 0,1

Attachment Time (by user) Minutes < 1

Removal Time (by user) Minutes < 1

Training Time (1st time) Hours < 1

Wear Time Without Discomfort Hours > 8

Device is Untethered Y/N Yes

Enclosure Temperature Degrees (Celsius) < 48.8

Tag Interrogation Frequency Seconds < 1

Minimum Tag Read Distance Feet 4

Team Roles

Magy Yasin ISEProject Manager, Ergonomics, Usability/Human Interface, Design

of Experiments, Maintenance and Manufacturing

Dave Taubman EELead EE/CE, RFID Reader Characterization, RFID Reader

Interface to MCU, RFID Antenna Interface

Curt Beard EEImpedance matching, Analog-to-Digital and Digital-to-Analog

Interface, Drive Circuitry, Device Characterization, Board Layout

Oliver Wing CEMCU Selection, Programming, Algorithm Implementation,

Memory Allocation, MCU Development Board Interface and Usage

Aalyia Shaukat EEVoltage and Current sense circuitry, Power Budget Analysis,

Battery Selection, Regulation Circuitry, Sensor Interface Circuitry

Stu Burgess MEAnalysis and Design of Enclosure: CAD, Heat Dissipation, Drop

Test, Manufacturing Enclosure, Cables & Connectors, User Input Device Selection, Attachment Ergonomics

Jeff Chiappone MELead ME/IE, Analysis and Design of Tactile Feedback:

Dynamics, Power, User Perception of Feedback, Cable Routing

Functional Decomposition

What How How How

Input Alternatives

Trackball-style Click wheel Touch screen

Keypad Knob

Input Alternatives – Pros/Cons

Pros Cons

Trackball Less moving parts, higher reliablity Too many “positions”, less intuitive

Click wheel Less moving parts, clicks become memorizableComplex design, may require  rebuild or reverse engineering

Touch screen Out-of-the-box functionality, can be made to order, re-programmable

May rely too heavily on “visualization” of gestures,etc.

Keypad Out-of-the-box functionality, can be made to order, waterproof, re-programmable

Some limit on size

Knob Simple to operate, higher reliability Very limited on size, less intuitive

Feedback Alternatives

ElectromagneticLevitation

Hobby Servo with Bracket

Temperature Linear Actuator

Vibrational Motor

Feedback Alternatives – Pros/Cons

Pros ConsElectromagnetic

LevitationIncredibly precise feedback, one moving part

Tech requires extensive research, parts not readily available, very expensive and labor-intensive

Hobby Servo Precise, steady motion Larger, heavier, expensive

Vibrational MotorOut-of-the-box functionality, inexpensive, tinyRequires damping to prevent housing from also vibrating

Temperature A unique experience, no moving parts May require steep learning curve

Linear Actuator PrecisionTend to come in larger sizes, higher weights, expensive

Power Supply/Charging

• What is the easiest way to charge the device?

• What is the most efficient way to power the device?

• Charging:o Wall adapter, Solar Cells, Kinetic Motion, USB

• Powering the Device:o Batteries, Other forms of stored energy

Charging – Wall Adapter

Pros• Dedicated port for charging• Not direction specific• Quick rate of charging

Cons• Additional Port needed to

charge device.o More holes in casing

• Need external wall adaptero Easy to misplace, bulky

Charging – Solar Cells

Pros• Would not need any external

power bricks• Can charge in any location

Cons• Needs to be sunny outside

o Unable to charge at night• Expensive to implement• Not enough current to power

the microcontroller

Charging – Kinetic Motion

Pros• Charge while you move• No external power brick

needed• Can charge in any location

Cons• Need to be moving to ensure

a charge on the deviceo Difficult to charge

• Bulky components needed to implement

Charging – USB

Pros• Fast charging• Uses existing port on device• Simple to implement• Can charge with computer or

with additional adapter• Universal, Inexpensive cable

Cons• Uni-Directional input• Need to be plugged into the

wall or computer

Powering the Device

• Must be easily rechargeable• Batteries have a large advantage over other

forms of power storage.o weight/size to storage density

• Several different types of batteries available

Batteries Available

• Li-Ion – Lithium Ion (80-90%)o Pros: Small, Light, No memory effect, low self dischargeo Cons: High internal resistance, Cell rupture if mishandled

 • Ni-Cd – Nickel Cadmium (70-90%)

o Pros: Can be fully discharged, high # Charge cycleso Cons: Expensive, Low energy density, Use of cadmium

 • Ni-MH – Nickel Metal Hydroxide (66%)

o Pros: High current drain, low resistanceo Cons: High output voltage, low capacity, long charge time

RFID Interfaces

• Skyetek-SkyeModule™ M9 RFID Reader • Available Interfaces:

o TTL (RS-232) – Requires Host Boardo SPIo I²Co USB – Requires Host Board

SPI – Serial Peripheral Interface

Pros• Simple• Can achieve high data rates• Simpler and more efficient in

point to point communicationo Lack of device addressing =

less overhead• Well documented interface

libraries

Cons• No flow control• Supports only one master

device• No hardware slave

acknowledgment

http://www.eetimes.com/discussion/beginner-s-corner/4023908/Introduction-to-Serial-Peripheral-Interface

I²C -  Inter-Integrated Circuit

Pros• Two wire connection• Good for >1 Slave device

Cons• Previous team was unable to

successfully implement design.

• Slaves need a unique address

Routing Algorithm

System Level Diagram

Sensor data

• The RFID reader can determine which tags are within range, but not their distances

• The RFID reader has a sleep mode that consumes less than 2% of the active power requirement

• Data from the compass and the accelerometer can be collected over time to determine a change in position

• Goal: use the accelerometers and the compass to provide most of the feedback, and use the RFID tags occasionally to verify the location

Map representation

• Major considerations include the size of the map file (limited memory space on the board), the ASCII character set requirement spec, and the 1000 maximum tags spec

• Maps consist of:o tags, each with an ID (12 bytes) and X and Y coordinates in

inches or centimeters (range: 0~4000);o map vertices (e.g., rooms, water fountains, bathrooms, hall

intersections), each with X and Y coordinates;o walking paths between vertices, which include the start and

end verticeso assumption that the device is currently only being used to

navigate one floor of one building

Map representation (continued)

• Possible solution: use base64 notation (ASCII-safe but still small)o Tags, with their large ID fields, will likely comprise the largest

part of the fileo With base64 notation, tag IDs require 16 'digits' and the

coordinates will use 2+2 'digits' = total of 20 bytes per tago The MCU memory must be able to accommodate up to 1000

tags = 20KB minimum per map (restricts MCU choice!)• Store the graph underlying the map of destinations and paths

densely (low connectedness)

Microcontroller selection

Selection: Arduino Nano (weak in clock speed, but sufficient in memory and I/O options, and can start programming quickly) with TI MSP430F5529 as a backup choice

Figure 1. Flowchart representing the Dijkstra algorithm. To convert to A*, changethe step that selects the vertex with the lowest distance value to instead pick

the lowest distance value/heuristic combination.

Morph Chart

Functions Options

Attach to user: location

Waist Neck Wrist Arm Hand Back Head Leg/ankle

Attach to user: method

Velcro Belt Watch band Spandex/sleeve Headband Backpack Glove  

Send output to user Lorentz Mag-LevVibrational motorPressure wristband

Servo spineCombo of

vibration & touch     

Receive user input Keypad KnobRetro-style

telephone dialTouch screen Scanner      

Power all components: charge

methodWall adapter Solar cells USB Kinetic motion        

Power all components: battery

Lithium Ion NiCad NiMH          

Microcontroller Arduino PIC TI OMAP         

Pugh Chart

Specs and Customer needs System I System II System III System IV

Datum

Charge time 0 - 0 -Non-visual commands 0 0 0 0Verification of user input provided 0 0 0 0Hands to carry system 0 0 0 0Hands to position/use system 0 0 0 0Attachment time 0 + + +Removal time 0 + + +Training time (input) 0 - + +Tag interrogation freq. 0 0 0 0Distinct inputs from user - - - -Noise at 3 ft. - 0 0 -Training time (output) - 0 0 +Size + + + -Weight + + + -Cost + - - -Device is untethered + + + +Number of 20' nav intervals on full charge + + + -Time to read map from electronic source + 0 + +Wrong command (direction) provided + + + +Distinct & distinguishable + 0 0 +Time between commands + 0 0 +Impact resistance + + - -Wear time without discomfort + - + -Doesn't draw attention + - + -

# of +'s 12 8 11 9 N/A# of -'s 3 5 3 9 N/A# of 0's 8 10 9 5 N/A

Total Score 9 3 8 0 N/A

System I System II Sytem III Sytem IV End configuration.:

Keypad Knob Touchscreen Scanner Keypad

Arduino PIC TI OMAP Arduino Arduino

Vib. Motor Pressure Wristband Pressure Points Vib. Motor/Touch Combo Vib. Motor

Li-Ion NiMH Li-Ion NiMH Li-Ion

Ankleband Headband Glove/arm Necklace Glove/Arm/Wrist

USB Power Solar Cell Wall Adapter Kinetic Motion USB Power

Input Decision: Keypad

• Size: LxWxD: 2” x 1.58” x .5” (smallest)• Weight .016 lbs/7.5g (smallest)• Manufacturer/Source: Sparkfun Electronics, adafruit.com,

futurlec.com, • Cost: $3.95 (cheapest) • Material: aluminum, plastic, silicone• Connection: 7 output pins • Meets Requirements: can be programmed to suit, some models

come with Arduino microcontroller support, requires only one hand to use

• Reliability: Low actuation force, waterproof, millions of actuations per life of unit

Pre-made vs. Build to Suit:

• Pre-made keypads: o May allow for design-specific layout, more intuitive to usero May be able to design a smaller keypad than most pre-made

models, but:o They may also have long lead times for manufacturers to

produce• Built to suit keypads:

o Require less time incorporating into team’s specific designo Have built-in Arduino support

Custom Keypads May Be:

Radial Directional

Functional and Ergonomic Slim and Compact

Custom Keypads Manufacturers:

• Top Bound USA• Baran• Miller Dial• Gray Hill – *very thin rubber keypads (.2”), also

sells individual keys with switches

Feedback Decision: Vibrating Motor• Size: 0.5” long• Weight .004 lbs/2 g (lightest)• Manufacturer/Source: Amazon, TrueSupplier, futurlec.com, • Cost: $2.99 (cheapest) • Material: aluminum, copper, synthetic foam• Connection: 2 copper leads• Meets Requirements: provides distinguishable output to user,

voltage allows for change in intensity, can be placed in multiple locations for diverse feedback communication

• Reliability: only one moving part, compact, suppliers guarantee millions of revolutions per life of unit

Decision - Charging Using USB

• The user will already have to be familiar with programming the device via USB

• Existing USB charging controllers are well documented and special IC’s already exist.

Decision - Li-ion Battery

• High capacity• Best size to weight ratio• Fast charging and long battery life• Charge lasts a long time

Example USB Li+ Charger

• Uses existing IC’s to implement and control charge and output voltageo Source: http://www.maxim-ic.com/app-notes/index.mvp/id/3241

Decision - SPI

• Pre-made functions and well documented library by Arduino

• Simple and easy to use for single Master-Slave configuration

• Previous team unable to implement I²C successfully

Questions?

Heat Dissipation

Assumptions• Housing is a rectangular box with four

equal sides• Bottom surface is well insulated• Uniform air temperature inside housing• Uniform temperature gradient on walls• Heat generated by all electrical

components lumped in to a single Qsource

QSideQTop

QSide

QSide

QSide

Qsource = 4*Qside + QTop

Conduction: Q = (k*A*∆T) / t

Convection: Q = (h*A*∆T)

Basic Equations

Additional Assumption• Qsource is floating within the box

Heat Dissipation – Initial Analysis

Calculations• Using a known ambient temperature and known Qsource temperatures can be

calculated by analyzing and summing up heat flow through each individual wall• Heat transfer through each wall is a simple 2D analysis• Convective coefficients will need to be estimated• Equations can be put in MatLab to quickly recalculate for different geometry/materials• Very conservative and simple analysis• If calculated temperatures are excessive a secondary analysis will be done

Tcomp Tair,i Twall,i Twall,o Tair,o

Qsource

Heat Dissipation – Secondary AnalysisAdditional Assumption• Qsource is attached to one of the walls• Designate sections of wall to only conductive or convective heat transfer

Calculations• Similar to previous analysis with addition of direct conduction of heat from the

components to the wall• If temperatures are still found to be excessive will need to consider use of more

conductive materials and/or incorporate heat sinks to reduce resistances in the thermal circuit

Tcomp Tair,i Twall,i Twall,o Tair,o

QsourceQwalls

Qcomponent conduction

Housing Material

Material Density(g/cm3)

Elastic Modulus(GPa)

Thermal Conductivity (W/(m*K))

Steel 7.8 200 52

Aluminum 2.7 69 210

Carbon / Epoxy 1.6 142, 10 ~1.5

E-Glass / Epoxy 2.1 45, 12 ~1.5

Polyester 1.4 3 ~0.2

• Results of thermal analysis, drop tests and projected budget will help determine whether to use metals, polymers, composites or a mix of materials

Human Factors•Interaction between user and product.    •Quick Response    •Constraints: to prevent mistakes being made    –Input      •Different sizes and textures•Input data twice    –Output•Feedback should be differentiable.    –Device shape: should help determine placement•Asymmetrical Shape

Adjustability

•Ensure product is secure–Position–Elastic material•Spandex–Adjust product first time to user–Release button •Ensure button is not easily pressed.

Testing - Types

• Drop Test (ME)• Thermal (ME)• Noise (ME)• Usability (ISE)• Comfort (ISE)• Attachment/Detachment (ISE)• Training Time (ISE)• Software Functionality (CE)• Battery (EE)• Amount of power used (EE)

Testing - User Interface

•Feedback Capability: Determine distance between vibrational motors–Order feedback parts–Determine minimum distance needed to sense different feedbacks•Input Error: Determine input error rate–Collect data on the number input error observed from sample

Testing - Methodology

Drop Test• Obtain sample sheets of possible materials

 • Cut sample to approximate size of one side of housing

 • Attach weight equal to projected weight of device to samples

 • Perform drop tests

Thermal Testing• Obtain thermocouples and heater capable of outputting power

equal to that of the device • Place heater inside manufactured housing unit and monitor

temperatures until steady state is achieved

Testing - Program

• Unit testing• Self-testing (power-on self test)• Algorithm correctness• Application testing

o Cooperation with sensor and feedback interfaceso Testing directional feedback on-site without mechanical parts

Use log files or USB communication to check resultso Comprehensive error cases – user input, sensor input, map

data

Risk AssessmentRisk Causes Effects L S I Preventative Action Owner

Losing saved work

Not using version control systems

Not creating backups in a separate location

Fall behind schedulePossibly forced to re-design 1.5 3 4.5 Use version control continuously

Make backups on other media Team members

Algorithm fails during user trial

Poor programmingNot enough consideration

for error cases

Device malfunctionsIncorrect signals cause discomfort

through directional feedback2 2 4

Test a wide range of plausible error cases, including hardware failure and disallowed user inputDocument program design and

code thoroughlyRedesign with simpler or more

deterministic algorithm

Oliver

Housing Breakage Material too weakMachining/Assembly errors

Re-evaluate material choicePartial/Full rebuild of housing 2 2 4

Order extra materialSimulate drop test with sample

materials before buildStu

Interfaces don't work as planned

Poor planningInadequate research

Inadequate programming ability

Forced to redesignFall behind schedule

More work during weeks 6-10Unhappy customer

2 2 4

Research thoroughly during design phase

Look for alternative interfaces in the same parts

Lead engineers

Hardware fails

OverheatingPhysical damage

Improper testing and handling

Replacement of parts shrinking budget 2 2 4

Order extras of fail-prone piecesKnow the parts' operating limits

Be nice to the deviceLead engineers

Parts arriving lateFaulty shipping method,

shipping scheduleloss of time, schedule 1 3 3 order early

each team member, according to duty

Risk Assessment

Schedule

Schedule Continued

Questions?