• Concept
• Motivation
• Competition
• Tech Specs
• Requirements
• Architecture
• Use Cases
• System States
• Risks
• Error Handling
• Implementation
• Test Cases
• Evaluation
• Lessons
• Fun Stuff
• References
• 18-549 Home

MEMBERS

Amber Imam Wai Yong Low Rashmi Sanbhadti Sudeep Yegnashankaran

And Navime!

PROJECT CONCEPT

The NaviMe system is meant to aid blind people navigate foreign buildings. When a blind person uses
a cellphone/PDA with the NaviMe software on it, the device will interface with the network as well
as a pre-loaded map of the building in order keep track of the blind person at all times, allowing
the device to guide him/her as well as provide helpful information regarding nearby points o
interests, such as locations of restrooms and elevators.

^ back to top

MOTIVATION

One of the scariest obstacles for a blind person is navigating a foreign place. With the use of the
NaviMe system, not only will the blind person feel more comfortable entering a foreign building,
but he/she will also receive live information about points of interests around him/her, such as
the location of nearby elevators, restrooms, receptionists, etc. Of course, such a device will
also for a sighted person trying to find his/her way around a new environment.

However, in simpler terms, blind people are awesome. That's really all the reason we need. Do you...disagree?!

^ back to top

COMPETITIVE ANALYSIS

Wayfinder Access GPS system for cellphone Not for buildings
Prague Royal Route developed by CUT Maps that help navigate through entire city (not buildings) with infrared technology Not made for cellphones or PDA's
Blip Systems Builds Bluetooth sensor networks for home/office usage Sell individual Bluetooth sensors that give off a unique ID and have built in power sockets Does not offer a navigation system Very expensive

TECHNICAL SPECIFICATIONS

• Nokia N95
• Microcontrollers with built-in Bluetooth,
  made by National Semiconductors
• Server to download building maps to pre-load device

• Assembly/C/C++ for core functionality
• Java for user interface
• Text-to-Speech software for communicating with the blind person

• Bluetooth to communicate with sensors
• Skyhook Wireless, WinPcap for using wireless routers
• GSM for using phone towers for triangulation

REQUIREMENTS

• Preloaded map of building
• Text-to-Speech software
• Give step by step directions to user in real-time
• Cheap, low power, maintainable infrastructure in buildings

ARCHITECTURE

^ back to top

USE CASES (INTERACTION DIAGRAMS)

^ back to top

SYSTEM STATES & TRANSITIONS

• Start Up
• Tear Down
• No Signal Case
• Navigation Case
• Obstacle Detection
• Downloading Map Information

RISKS & MITIGATION STRATEGIES

• Hard to coordinate WiFi/bluetooth sensors - Latency, varying signal strength
  • Can write robust code to account some variation/errors
  • Pairing of WiFi and Bluetooth will allow for some error correction
• Reliability Issues - Routers/nodes might go down
  • Hardware redundancy
• Maintainability of Bluetooth network
  • Optimize on power usage of the Bluetooth network
  • Possibly have give them dedicated power lines
• Temporary Obstacles that change the map - Spills, out-of-order elevators
  • Routing algorithm will recalculate route
• Require special training for building admins to be able to update databases/map
  • Make database/map interface user friendly
• Accuracy
  • Hardware redundancy

ERROR HANDLING

To be completed.

IMPLEMENTATION DETAILS

• Image of midsemester app. Detected nearby wireless routers and displayed their MAC addresses and signal strength.
• Fingerprinting
  • Fingerprinting relies on a training phase in which a radio map of the environment of interest is constructed by taking a series of radio measurements in multiple locations. A measurement records the strength at which signals emanating from a group of radio sources are heard at a given location. Once the training phase is complete, a client can estimate its location by matching the current measurement to the set of measurements collected in the training phase. "A. Varshavsky et al. / Pervasive and Mobile Computing 3 (2007) 698Ð720"
  • Create a database of fingerprints ( list of surrounding mac addresses and their corresponding signal strengths )
• Convert blueprint of map to graph with nodes and edges
  • Use Dijkstra's algorithm to find the optimal path from start node to destination node
  • Nodes
    • Center of rooms
    • Space in front of doors (in hallway)
• Use localization algorithms to calculate user's position
  • Use K-nearest neighbors to find a matching fingerprint in the database
  • Use user's current position to direct him/her to the next nearest node in the path
• If time permits, recalculate route if user deviates from original path

  TEST CASES

  Testing our architecture
  1)

  • Test effect of time of day on wifi signal strength (reliability)
  • Keep location constant, vary time of day
  • How: Have a person start at one end of the hall, with the end of the hall being the destination. Once the phone starts giving directions, the user should start walking and follow the path indicated.
  
  2)
  • Test range of the gumstix bluetooth antenna (reliability)
  • Keep location and amount of foil constant, vary time of day
  • Keep time of day and location constant, vary amount of foil
  
  3)
  • Test accuracy of identified location (functionality)
  • Static: keep time of day constant, vary location
  • Moving: keep time of day and path constant, vary speed
  
  Testing our application
  1)
  • Name: Hallway Location
  • Why: It has the very basic building blocks of a generic layout and has the minimal functionality that we should be able to attain.
  • How: Have a person start at one end of the hall, with the end of the hall being the destination. Once the phone starts giving directions, the user should start walking and follow the path indicated.
  • Fault-free
  
  2)
  • Name: Corner Location
  • Why: It has the very basic building blocks of a generic layout and has the minimal functionality that we should be able to attain.
  • How: Have a person start at one end of the hall, with the end of the hall around the corner being the destination. Once the phone starts giving directions, the user should start walking and follow the path indicated.
  • Fault-free
  
  3)
  • Name: Lobby Location
  • Why: This situation is a major source of problem for the blind and therefore is a minimal functionality that we should be able to attain.
  • How: Have a person start at the Market Street Deli/far end of Newell Simon Atrium, and direct the person to the roboreceptionist.
  • Fault-free
  
  4)
  • Name: No wireless routers
  • Why: The user may enter an area that has no available wireless router signals.
  • How: Disable wireless on the phone to simulate the condition where there will be no router signals available.
  • Fault-injected
  
  5)
  • Name: No map
  • Why: If the user tries to access this application in an area where a map is not available, we need to be able to take this into account.
  • How: Since we will be using Newell Simon as our test area, any area other than this building will not have a map available. Have the person go to another building at CMU other than Newell Simon and try to access our application.
  • Fault-injected
  
  

  EXPERIMENTAL EVALUATION

  To be completed.

  LESSONS LEARNED

  To be completed.

  FUN STUFF

  Random pictures :-D

  See, it's 'Shmi!	Brrrr it's cold!	Eep eep!	No way, it's Wai!

  
  

  The NaviMe Team...Brady Bunch Style!

  
  

  Trying to evaluate the use of Aluminum foil to reduce the range of the Gumstix Bluetooth antenna.	Oh where, oh where could that Gumstix be? Oh where, oh where could it be???

  
  

  Hey buddy, what's in the box???	You can't have what's in the box! It's MIIINE!!!

  REFERENCES

  • Dan Rossi
  • Project Proposal and Requirements, Team Project Presentation, February 1, 2008
  • Design & Architecture, Team Project Presentation, February 15, 2008
  • Mid-semester Project Status, Team Project Presentation, Date TBD
  • Test Plan & Experimental Validation Team Project Presentation, March 28, 2008
  • Final Project Presentation, Team Project Presentation, Date TBD
  • Project Poster, May 2, 2008
  • Demo Video, May 2, 2008

  ---

  Back to the top of this page
  18-549 course home page