• Owner : Nirmal
• Description : Explore various interaction techniques for a circular touchscreen on a wrist watch. This will be a two part project.
  • Create a collection of interaction techniques for a variety of common interface tasks. Examples include scrolling, selecting from a list, text entry, value sliders…
  • Perform task analysis to determine the most useful tasks which can be completed from the watch
  • Create a UI to demonstrate a cohesive watch based user experience.

• Owner : Dan
• Description : How can we do touchscreen interfaces on wristwatches? Your finger is very large compared to the screen, so what techniques can be used to help the interaction? This project will follow up a Fitts Law study we've been working on this semester by investigating the effects of different kinds of feedback on the accuracy and speed of touchscreen interaction. Skills needed for this project include Python programming and potentially some electronics experience.

• Owner : Christopher Skeels
• Description : See this UIST 2007 Extended Abstract christopherskeelsuist2007posterextendedabstract.pdf and this rough video for what has been done. The following are projects you could do to build on the existing technology:
  • Interface & Visualization Improvements, e.g., Are the visualization colors correct/intuitive?
  • User Study, e.g., Compare a group of users sculpting with system to a group of users sculpting without the system

• Owner : Harley
• Description : Create a mobile phone based game that will help children learn how to finger spell.

• Owner : Harley
• Description : Create a mobile application that allows a hearing parent to find videos of signs they want to make to a deaf child. Initially develop system for relaying spoken words to a person who will in turn relay the videos back to the parent.Evaluate system to determine the usefulness of such a learning tool. To see more sign language projects and tools go tohttp://www-static.cc.gatech.edu/program/cats

• Owner : Val
• Description : Evaluate multiple design solutions for a mobile deaf-to-hearing communication device. Current solutions are using pen/paper, typing, phrasebooks and sign-to-speech technology. Use a Wizard of Oz system to simulate sign-to-speech technology.

• Owner : Val
• Description : Develop web based learning tool for ASL. Study effects of vocabulary word spacing (see Mnemosyne, http://www.supermemo.com/english/ol/sm2.htm). Develop mobile phone version of the same learning tool.

• Owner : Tracy
• Description : Create new toys for young children to play with. They will house various sensors (accelerometers, GSR, …) Come up with ways to sense new things (i.e. use conductive fabric to build a mouthing detector)

• Owner : Tracy / David
• Description : Apply Experience Buffers technique along with activity recognition code to find interesting patterns in sensor data collected from babies.

• Owner : James
• Description : Create a jewelry ensemble where all the pieces are aware of each other. If they are separated have them alert the wearer.
  • Could potentially use the wearer's body to conduct presence messages instead of a wireless protocol that would require separating out people.

• Owner : James
• Description : Build a functioning bluetooth headset into a pair of earrings.

• Owner : James
• Description : Create clothing that somehow communicates the state of the user. How interruptible are they? Are they angry?

• Owner : Thad
• Description : Make a tie and embed some electronics. Ask Thad for details. This has some serious demo/outside appeal possibilities.

• Owner : Travis
• Description :

Last time this course was offered, my group used piezoelectric sensors on the wrist/ankle to classify gestures for application control (see more about Hambone).

I believe there are other possibilities for piezoelectric sensing systems. One example is to use time-of-arrival and triangulation of pulses to detect tap location on various surfaces (such as a tabletop, on glass, etc). Thus, it might be possible to create a wearable, “take-anywhere” keyboard.

The 1st project could focus on characterization, and a possible 2nd project could focus on implementation. This project requires minimal hardware experience, but should have a good handle on signal processing (preferably MatLab – though python would probably work).

If interested, please see me (currently located near TSRB 240) or email me at this address.

• Owner : Travis
• Description :

International Fashion Machines makes some very simple, fabric-based electronic switches/buttons (you can read more about it at Hizook).

The hardware is pretty simple (some conductive fabric and some cheap Q-Prox ICs). From the hardware side, it should be fun to experiment with resolution, capabilities, and device control (and PCB fabrication, oscilloscope use, etc). From the industrial design side, this should be very simple to integrate into clothing. Thus, this would be a great multi-disciplinary team project.

It would be best if the hardware person had some hardware experience – soldering, breadboarding, microcontrollers experience would definitely be helpful…

If interested, please see me (currently located near TSRB 240) or email me at this address.

• Owner : James
• Description : Automatic Whiteout is a system which automatically corrects common mistakes made on mini qwerty keyboards. Upgrade automatic whiteout to use more timing information from existing typing data. Evaluate improvements to typing speed/accuracy.

• Owner : James
• Description : Port Automatic Whiteout algorithm and application to J2ME phones for evaluation of algorithm on an actual mobile device.

• Owner : James
• Description : Twidor was a typing tutor for the Twiddler keyboard. However, it can be used with other keyboards. Port Twidor to the iPhone to determine the learning curve of the iPhone soft keyboard.

• Owner : Dan
• Description : The MAGIC system helps a user pick new gestures for use in mobile computing systems. It does this by comparing a candidate gesture with a database of the user's gestures taken over a week or month. This project will involve implementing new functionality in MAGIC and building applications with it. Python and C skills are required, as is familiarity with machine learning and/or pattern recognition concepts.

• Owner : Nirmal
• Description : For a summer internship project at Motorola I created a mobile version of DDR. Unlike the standard DDR where you step in the direction of the arrow, in my version you kick in the direction of the arrow. See the description here and a demo video here.
  • For this class, I'd like to have the game judgement improved. As of right now it uses simple thresholding to determine if a kick has been made. However, this is not a very accurate way to judge and give feedback to the player.
  • Additionally, we would like to have multiple players in a co-op mode (line dancing) or a competitive mode (you got served).

• Owner : David
• Description : Test activity recognition code using data sampled from playing WiiSports. Attempt to reconstruct activities in the game given raw data and validate against ground truth.
  • The original path for this project was to use a computer between the Wiimote and Wii to capture, log and then relay the events. I don't think anyone has successfully faked a WiiMote from the perspective of Wii. See WiiMote Drivers
  • Another option is to use the Macromedia Flash to create a custom game. This actually makes it easier to both log data and capture ground truth simultaneously. The harder part then becomes to create a compelling game. See WiiMote Flash library

• Owner : Jung-Soo
• Description : Create tools to make GART easier to use for outsiders. Add things like leave-one-out testing, cross validation, data visualization for sensor streams, data management tools.

• Owner : Jung-Soo
• Description : Update GART to be pure Java. Remove dependency on HTK and external bluetooth stacks. Make GART work under Mono so that it can be used in other languages such as Python.

• Owner : Nirmal
• Description : Replace my existing gear shift with one that has embedded proximity sensors. Awhile back, I did a demo for Toyota in my car ('07 Camry) that basically involved using the AUX input on the stereo and a rear-view mirror mounted proximity sensor to control voicemail message sending. Basically, we used varying numbers of fingers to represent different commands.
  • Mounting the sensor on the rearview mirror is not ideal. If you move the wrong finger across it you are going to upset someone.
  • Mount multiple sensors across the top so we can determine direction of movement.
  • Mount one on the left side so when all our covered you know that I'm changing gears and not gesturing.

• Owner : Dan
• Description : The Twiddler is a one-handed chording keyboard used by Thad and many other wearable computer users. Unfortunately, the Twiddler is no longer being sold, so we need a new one. This project will involve working on a new Twiddler. There are several potential aspects to this project, depending on your skills:
• electronics knowledge or interest
  • make a new keyboard board for the Twiddler case, including buttons and logic to allow the board to be plugged directly into USB (or to function over Bluetooth).
  • take apart an existing Bluetooth keyboard and use its chip to build a new Twiddler board
  • put a IRDA transceiver into the Twiddler so two users can transmit text back and forth; this could work like a “universal” clipboard, where user A copies the text and user B can paste it.
  • add capacitive sensing to the keys of the Twiddler to allow “previews” of functionality for tutorials, or to the body of the Twiddler so the computer can detect when the Twiddler is being held.
• 3d modeling knowledge or interest: model the existing case or create a new case for the Twiddler board; then print it on the 3D printer and see how it works
• software: make universal tab-completion for your OS of choice. This would work just like tab completion in IDEs, or the automatic word completion on the iPhone, but in all programs. One potential way (under X) is to hook into the foreign alphabet typing (like Japanese) functionality to listen to and replace keystrokes.

• Owner : James
• Description : Write software for dual camera phones that uses the inward facing camera to track the user's eyes. Use this information to create a novel interaction based on the user's attention.
  • See: eyeLook

• Owner : James
• Description : Develop a quick communication tool for mobile devices that are within some range of each other.
  • See the Unix talk utility: “transmitted each character as it was typed, lending a more immediate feel to the discussion than recent instant messaging clients or IRC.”
  • Possibly deploy at ETRI show if we use J2ME.

• Owner : Nirmal
• Description : When you touch two fingers to a single point touchscreen the outcome is a point that is on the line segment created by the two touch points. By controlling the relative pressure on these two points you can move it back and forth. This is now effectively a 1-D touch screen cursor which is not occluded by the finger. What does it mean to use this form of interaction on a mobile device?

• Owner : Nirmal
• Description : Use standard single point touch screen to emulate some multi-touch style interactions. Create a few demo apps (2 finger scrolling). Involves some simple pattern recognition.

• Owner : Nirmal
• Description : Use ultrasonic pings from a cellphone to determine what type of surface it is near. Is it in the front pocket or back? On my desk table? Conference table? Could I capture sounds to implement a tapping interface?

• Owner : Nirmal
• Description : Use standard bluetooth phones to interact with a projected display. Explore possibilities of using viewports on phone to determine context. How do you establish control focus? Possible demonstrations could include card games (poker, rummy…)

• Owner : Kent
• Description : Current syncing practices between our myriad of devices is broken. Why are my email programs so disjoint? Work with Kent to develop an architecture (possibly at the toolkit layer) to help solve this problem

• Owner : Nirmal
• Description : Anyone seen Big? Tom Hanks dances on a giant piano in a department store. Those things are cumbersome. What I want to do is get some carpet tiles from Home Depot, stick some RFID tags underneath and then use a foot mounted RFID reader to recreate the piano. This way to move the piano, you just stack up the carpet tiles and take them with you.

• Owner : Dan
• Description : The Firefox spellcheck dictionary doesn't have common words like “USB” in it. Write a Firefox extension to connect to other peoples' dictionaries and use them as well. Take into account people marking misspelled words as correct and attacks/pranks (like several people adding “teh” as a correct word). Potentially expand to include reputation scoring.

• Owner : Travis & Matt
• Description :

We have built a device (a surface) capable of transferring power via inductive coupling for short-range (less than 15cm) wireless power transfer. There has been a lot of 'buzz' about similar projects in the press, but no systems are readily commercially available.

We have conceived of a few applications. One example is a table where any electronic gadget (for example, a cell phone or laptop) placed on the table automatically recharges. We are searching for other compelling applications, and this is a chance to develop wirelessly-powered devices before such systems become widely commercially available!

I don't want to discuss my particular application of the surface (at least not yet, since it is in the publication pipeline), but I would be happy to give demonstrations. If interested, please see me (currently located near TSRB 240) or email me at this address.

• Owner : Jay Summet
• Description :

DigiQuilt is a computer game for teaching children shapes, colors, and fractions. In this project, you would build a physical DigiQuilt gameboard and pieces using the GVU Laser cutter, and use computer vision libraries (ARToolkit, OpenCV) to detect and track the gameboard, as well as identifying the colors of pieces on the gameboard.

• Owner : Jay Summet
• Description :

You would design and assemble a 3-axis computer controlled mill for rapid fabrication. Servos and drive electronics are provided, as well as an initial platform.

• Owner : Ed Price
• Description : We have developed a prototype system to use a simple augmentation to the power system (one device plugged into a outlet) to track people with active badges throughout the home.

• Owner : Ed Price
• Description : How can we detect events and activity in a home by the detection of transitory events on the power lines in a home. Detect power loads being drawn throughout the house to determine energy use. Simple systems to inform residents on the sustainability of energy choices.

• Owner : Ed Price
• Description : A consortium of 150 or so manufacturers of connected home products (everything from typical smart home systems, lighting systems, intercoms, garage door openers, etc) using a ad-hoc mesh network. How can we update the Aware Home to take advantage of these mesh networks connecting smart devices throughout the home.