I'm trying to build a smart frisbee for use in Ultimate. I'd like to be able to track the position of the frisbee in a playing field with pretty good accuracy (within ~2ft). GPS is a possibility, but I'm not sure if it is accurate enough. Are there any other technologies that would help, such as some system where I set up transmitters on the four corners of the field and place a receiver on a frisbee?
You may be able to do this with a TrackR like low energy bluetooth tag attached to the center of the frisbee, and then four readers or more readers, along the side lines. The readers would then use the power level and triangulation to determine the position of the frisbee.
GPS has a relatively low accuracy for every sample (in the order of 5-10 meters 2sigma [95% of values]) at 10Hz. A frisbee is not slow, so averaging (which reduces the position fix rate) to reduce the confidence interval is not going to make it suitable.
Inertial measurement units are much faster and much more accurate but drift over time (also called "random walk", due to the double integration of noise on accelerations). It may be acceptable for you, especially if you reset them regularly using an absolute position sensor or by putting the frisbee at a location which position is precisely known. Data fusion is doing this automatically on-the-fly, that's what GPS navigators do on top of shifting the position to the nearest known road.
For accurate absolute measurement, you can use differential GPS (also known as d-GPS) which will require a fixed beacon somewhere on the field and a wireless connection. This can give sub-meter accuracy down to 15cm.
Though d-GPS is perfect for you, it's usually complex and/or expensive, so I'd recommend turning yourself to IMUs with at least 6 degrees of freedom. You may want to decouple the system from the frisbee spin (i.e. bearing), in which case you'll need an IMU which integrates a magnetometer to take care of the angular motion of the accelerometers frame of reference while decoupled.
GPS is a good bet, both in ability to implement and low cost to experiment with. Plus a robust frisbee system there could be marketed to disc golf players. It will suffer the precision issues you mentioned.
Likely the best solution is computer vision. It won't be easy. If you're willing to add electronics to the disc you may as well place UV LEDs on it and track it with a Wii type system in addition to standard video tracking.
A microcomputer with an accelerometer and wireless transmitter on the frisbee can calculate and send its 3d position. I think it is possible to construct them sufficient in weight, except the battery, I'm not sure about it.