CSE477 Project Ideas
Organization for the projects
Projects will be organized around a few themes. Each group will consist
of two students. However, several groups may end up working very closely
on projects that fit tightly together. Hopefully we'll have every group
also be part of a cluster around one of this quarter's themes. Coordinating
projects that share a theme will give you an experience of working in a
larger team while still being able to focus on the work of your group of
two. I'm also hoping to have at least one manager for each theme or even
a manager for specific projects. A manager's job is to meet with the group(s)
weekly and provide suggestions and guidance to keep the project on track
much as an industrial supervisor might do. I have at least one external
manager, John Wedgwood, who has extensive experience in the embedded systems
industry, and several internal managers, my graduate students, who are
doing projects and research in very closely related areas to the themes
of this course's projects.
The themes for the projects this fall will be:
-
exploting geographic information systems,
-
home automation, and
-
body-area networking.
These are by no means unrelated. In fact, the hope is that as many projects
as possible can be related to each other. The choice of these domains
corresponds to some areas of active research in embedded systems, user
interfaces, networking, and web-based services. By focusing on these domains
we hope to better leverage the work of the class to provide a more integrated
experience (in dealing with a range of related projects), work in larger
teams, develop some infrastructure for the research program, and provide
a more coherent story to tell about the projects (which will be translated
into video form after the quarter is over).
Each group will have a "manager" that will be responsible for providing
a sounding board, advice, and a constant pressure to make progress.
John Wedgwood will serve as such as a manager. John has extensive
experience in the area of embedded systems and in industry management.
Some of my graduate students, working on the Portolano
Expedition, will also act as managers if there is a close match between
a course project and their research interests. We may also have other
managers depending on which projects are chosen by the class.
Some examples of devices/services
Lets start by providing some example computing appliances that would fall
in each of these domains.
Home automation
There has been lots of activity recently by companies marketing home automation
products. Most of these involve a PC tied to a local area network
of some kind. Often the network is based on the power or telephone
lines. Two developments are worthy of note: 1) the use of embedded
web servers to give users control of their home from a remote site and
2) the development of wireless RF technologies for connecting devices.
However, most of the applications are based on command style interfaces
where the user specifically controls the home environment. We would
like to investigate more autonomous home control systems that do things
based on knowledge of the users' whereabouts and intentions. For
example, in the operation of a home heating subsystem for power efficiency
with maximum comfort.
emWare embedded web servers
Hydra
embedded web servers
HomeRF networking
Body-area networks
A new model is emerging for task-specific devices that relies on a body-area-server
and an entire constellation of devices that communicate with it and through
it to the greater web. New development in "through-the-skin" networking
is making it possible to create a new set of devices that communicate only
when the user is touching (or physically very close to) them. This
style of networking enhances privacy and makes devices "borrowable".
One can now pick up someone else's PDA and have the other person's data
become inaccessible (because they are no longer touching it) and have the
new holder's data get downloaded to it through that person's body-area-server.
Many new devices (e.g., pedometer, pointers, preference holders, etc.)
are now possible.
IrDA
Bluetooth
RF Monolithics
RF Microdevices
Geographic Information Systems
There is an ever-increasing amount of GIS data becoming available.
Emergency services (EMT, fire, police) would like to exploit this data
when they reach the scene of an emergency. Ideally, officers would
have maps of the area, the interior floorplans of buildings, and maybe
even the position of persons available as they reach the scene and then
have constantly updated on the scene. Users a museum or campus would
like to have detailed information about how to find what they are seeking.
Travelers would like to have always accurate information about the traffic
they can expect, flight delays, etc..
Project ideas to serve as starting points for thought (in no particular
order):
-
Virtual Neighbor: there are a set of devices commonly found in a
home or easily installed that could help people keep a virtual eye on an
elderly relative. For example, motion sensors, door/window sensors,
thermostats, and microphones could be used to do what was done traditionally
by the friendly neighbor. Unusual situations, for example, lack of
motion or a door left open during the night, could be brought to the attention
of interested parties so they could take appropriate and timely action.
The interesting part of this project is to create an architecture that
makes it trivial to add/remove sensors and provide a smoothly varying quality-of-service.
-
Location-Tracker: use a combination of GPS (for outdoors) and RF
(for indoors) systems to keep track of people and objects. This information
could then be used by applications to deliver specialized content or dialogs
to the user that are dependent on the location. A simple application
of this would be to locate the nearest printer to my current location.
-
Magic Lens: instrument a PDA with a compass and accelerometers so
that an application can be aware of what direction it is currently facing
and its angles of inclination. This information, coupled with location,
can then be used to determine what the PDA is in front of and specialized
content can be displayed on the screen thus serving as a magic lens.
One could adapt a camera to display an image on the PDA screen so that
it appears to be see-through but with information overlayed on top of the
image.
-
Campus Navigator: develop an interface to on-campus GIS to deliver
information about buildings and facilities. The idea it to be able
to show someone, as they enter or face a campus building, where the important
offices, receptionists, and restrooms are located. If someone has
an appointment in the building, then also indicate where there destination
is located and how to reach it.
-
Itinerary Planner: keep track of where a user is scheduled to be
throughout a day and monitor conditions so that itinerary adjustments can
be made in a timely manner. For example, a business traveler may
want to be conscious of traffic delays to the airport, flight delays, changes
in others' appointments.
-
Migrating Audio: have audio follow a person around an office or
from room to room in a home. Use available speaker services to keep
the volume as even as possible by using multiple sources and fading from
one set of speakers to the next. This is, of course, coupled
with a location-tracking system. The difficulty is increased dramatically
if we consider multiple people each with their own audio stream interacting.
-
Police/Fire/EMT Mapping: emergency service workers would like to
receive maps of their destination prior to when they arrive on the scene
including not only street maps but also building floorplans and, possibly,
locations of people within the buildings. For example, fire departments
may want to see where people are concentrated when they arrive at a fire
scene, EMT workers may want to follow a tag right to the patient needing
care.
-
Body-Area-Server: a gateway between devices carried by a person
and the wired information infrastructure. It would have a wireless
LAN connection as well as the body-area-network connection. The challenge
is devise an architecture where devices can be easily picked up and used,
computation can be off-loaded from the device, and communication with the
wired web can be optimized.
-
Digital Pill Box: keeps track of when a user needs refills (ordering
them automatically) and when a user opens it to dispense prescriptions.
Whether patients actually take their medications is crucial information
for physician who may otherwise mistakenly increase dosages or prescribe
different medicines.
-
HP Badge: we possess a couple of StrongARM based "badges" from HP
that include a collection of sensors. This device could be used as
the body-area-server or as a way to prototype devices. The badge
requires some software development as well as looking into a new operating
system such as Linux.
-
Measuring Body Motions: an alternative approach to monitoring people
with cameras, is to use accelerometers and distance measurement devices
(e.g., ultrasonic range-finders). This would permit investigation
of user interfaces that rely on gestures and movements without worrying
about obstructions and camera angles.
-
Automatic Appliance Control: build on a project from last year that
created a modular system for deploying appliances around a home and have
them automatically integrated into the home automation control system.
User interfaces for the appliances are automatically downloaded from the
web and rooms are outfitted with IR beacons that broadcast a unique identifier
for each room.
-
Death to buttons: many consumer products are encrusted with buttons.
This is particularly annoying for small devices (e.g. calculator-watches,
pagers, remote controls). Buttons are only one of many ways you could signal
an operation to a device. In this project, you'll take the button-encrusted
device of your choice, and explore alternate input modalities to off-load
some of the button functions. For example, using various types of sensors
(pressure, tilt, accelaration, heat), you could have the device sense when
it is squeezed, tilted, stroked, grasped, shaken, held, and so forth, and
have some of those gestures map to some of the button functionality. Some
of the questions for you to address are: which button-encrusted devices
are best suited to this? Which functions do you choose to offload? What
alternate gesture replaces the "press a button" gesture for those functions?
How do you distinguish between deliberate and accidental gestures? What
feedback mechanism do you use?
-
Preferences Holder: a key chain size device that can hold personal
preferences for home and auto configurable appliances (e.g., temperature
at home, radio settings in car). It would communicate via RF wireless
(short-range) when you enter a door or turn the key in the ignition.
It would be updated automatically when appliance is manually adjusted and/or
through a UI on a PC or PDA. Could also keep a log of an automobile's
state (e.g., next service, fuel level, location, etc.).
-
Day Tracker: use GPS and other data (e.g., known times of wireless
and wired connections) to build up a map of a day's movements.
-
Body Movement Tracker: use accelerometers positioned on parts of
the body and connected to a PDA via wireless links to keep track of what
a person does throughout the day. This could include walking time,
exercising time, gestures, etc..
-
Haptic Joystick: work with Prof. Hannaford in EE to apply the Microsoft
Haptic Joystick (that his group has already reversed engineered) to use
as an I/O device for a PDA or other portable application.
-
Assistant for the Blind: add a transceiver to a Braille display
so that a blind person can find out about the different objects/people
they may encounter during the day. The display would be belt or arm
worn. An example may be a web client that adapts web data for display
to the blind.
-
Bitmap Capture: often one would like to capture a text fragment
from a magazine or newspaper. Ideally, a PDA could have an optical scanner
in a corner of its case, capture the bitmap, and process it using an optical
character recognition (OCR) algorithm. Due to its current limitations,
an approach to prototyping such a feature would be to connect a commercial
pen-scanner to a PDA port and collect the bitmap into a form. When the
PDA's data is hot-synced to a PC, the OCR software could be run there and
a new translated record created. Both pen-based scanners and PC OCR software
exist today and could be purchased to complete this project.
-
PDA Chording Keyboards: it would seem to that there is a middle
ground between pen-based Graffiti and hunt-and-peck QWERTY keyboards for
entering text. A set of switches under the users fingertips could be used
to construct an optimized chording keyboard that presents a quick menu
of choices for the user to select with the stylus. Huffman coding may be
a way to start. The screen space occupied by such a menu would be much
smaller than a keyboard and could be less distracting during use in an
application.
-
Hypermail Subscription Service: it would be useful to have a set
of hypermail-boxes where different announcements (colloquia, course schedule,
faculty/staff out-of-town, visitors schedules, etc.) could be sent. The
Pilot application would allow a user to subscribe to any subset and automatically
download new message data (and remove old) whenever the user hot-syncs.
Some of these would require new applications (e.g., visitors schedule)
for viewing, other could be integrated with existing applications (e.g.,
colloquia - with abstracts - into date book)..
Comments to: cse477-webmaster@cs.washington.edu
(Last Update: )