I have a class called Senior Design next semester and we are looking for fun or interesting ideas that anyone might have for an embedded project. We will have Electrical Engineers, Computer Sciene, and Software Engineering majors working on this.
All suggestions are welcome. Some ideas we are thinking about now are:
How about a program that can survey a building from photos. It takes in photos (interior, exterior or both) and spits out CAD files.
(The same technology might be useful for land surveys: input photos, output DEM and other GIS data.)
For the electrical engineers, a handheld laser device that measures the dimensions of a room would be cool. (I'm sure there's electrical issues there!)
I'm interested in precise, quickly derived CAD models of houses because you could then do thermodynamics and other physical modeling to figure out how to make the building more energy efficient.
Many cool things to do with cell phones, in particular of the Android type!
On the general theme of having a device of sorts (either attached to the cell phone, not) that "does stuff" as directed / controlled by actions on/to the phone.
Because of the relatively short design and implementation period, and the need to coordinate between between EE and CS, the phone idea is also good, for it allows the CS folks to start coding and testing elements associated with the phone, even if the "device" isn't there yet.
Although, the new capabilities of phones (GPS, embedded camera, fancy screen and input devices...) provide fertile ground for fun, the idea to "hack" into the heap of discarded phones of the previous generation, a bit à la "Junkyard Wars" could be fun too.
Also, many cheap toys include reusable servos, motors and such.
You know the DARPA challenge with autonomous cars? How about a cheaper, easier version with radio controlled toy cars, controlled from a laptop? You could lay out a course on a gym floor and have the car find its way through it. You could make the course as easy or as hard as appropriate, once you get the car basically working. Or even have two cars race.
How about something that can modulate the speed of a computer's cooling fan to allow it to produce an audible beat frequency allowing brainwave entrainment for better concentration. See NeuroProgrammer2 for an example of the uses for this kind of thing.
You might need to be able to control several computers to acheive the effect, and for it to be audible to everyone in an open plan office...
(one side benefit will be that, as you're nearing completion of your project, you should find yourselves with more rather than less mental stamina!
Two magazines that might give you some ideas are:
Circuit Cellar Ink, and
Elektor.
http://www.circuitcellar.com/
http://www.elektor-electronics.co.uk/
If nothing else, look at the adverts, as you'll probably see things to get your mental juices going.
Failing that, here are some ideas:
Idea 1: An NTP server with GPS receiver. GPS is essentially a glorified clock, sending out signals that are basically "at the sound of the tone, the time will be..." and calculating from several of those signals where one would have to be, for all of those signals/waves/messages to be true. If you don't care about position and velocity, then you can get by with a single receiver. Microwaves, spread spectrum, kalman filtering. Broadcom might let you have some eval chips, I don't know if they do eval kits anymore.
NTP is a protocol to help computers synchronize. It handles a number of issues that most folks don't know or care about. Yet if you're, say, selling stocks, when did order_X come it? [1] Many PC clocks vary by a lot, and some EQ bugs with AAs (message being something like "you must wait 98290528592034 minutes before you can use that again") were caused by players not setting their PC clocks to the same day/month/year as Sony's servers.
Possible market: there are a number of these on the market at this time. Most sell for thousands of dollars. Get it down to a couple hundred dollars and you could probably sell lots to corporations for about $300-500.
Idea 2: Procedural [2] brass instrument. Your engineering library should have this book. Chapter 7 talks about physical modelling of musical instruments using various algorithms. Figure 7.9 of the book (p287) is similar to fig 1 of this document [3]. Cook wrote his dissertation on some stuff like this, and is cited a bunch [4].
Other oddball links and stuffs:
http://www.music.mcgill.ca/~gary/ThePipe.html
http://ccrma.stanford.edu/marl/
http://www.music.mcgill.ca/~gary/ (if you're interested in procedural woodwind, see this guy's dissertation)
http://www.musickit.org/
http://ccrma.stanford.edu/~jos/swgt/swgt.html
Notes and digressions:
1 - Ancient chinese saying: man with one clock know what time it is, man with two, never sure.
2 - By procedural, I mean using physics (mostly evil math like 2-D transmission lines for wind/brass instruments) to model the sound of things, even things that can't be built. Plucked
3 - A maraca controller has actually been used for the Dreamcast and Wii for a single game: Samba de Amigo.
4 - I'd like a copy of it, but the only place that appears to have it online is ACM and I let my membership lapse (again). It is reference 6 in the paper just prior to note 3.
For my senior embedded project, we worked on Sumo Bots. In our case the bots were prebuilt, but it was amazing how much strategy you could pack into an embedded system. I got bonus points for designing a state based machine.
The robots were very simple, two big wheels attached to a frame, with some weight keeping the front low to the ground. There were also a few sensors: a distance based sensor, an eye that could see the white lines on the outside of the ring, and one touch bumper.
The competitive aspect of wrestling robots really got our juices flowing. You could do something similar with racing robots, climbing robots, rally robots (darpa), or even tug-of-war bots.