18. Java-Based Kalman Filter Learning Tool

Prof. Greg Welch (welch@cs.unc.edu, 962-1819)

Abstract

The Kalman filter is a set of equations used to recursively estimate the "state" of a process (the position and orientation of an airplane for example) given a time-varying sequence of noisy measurement (e.g., air speed, pressure, temperature, engine thrust, etc.) It is a popular mathematical estimator for a variety of reasons. It is optimal onder certain conditions (rarely met but often close), it is very efficient, and very robust.

I maintain a relatively well-known web page on the topic, with papers and pointers. I would love to have a simple Java applet (or whatever) that operates from with the web browser, and gives a user the ability to change various parameters and see how the Kalman filter would respond for a given (randomly generated?) set of noisy measurements. You could be famous!

Implementation Notes

Should work with current Mac, PC, and Unix browsers.

Related Links

Back to Project Candidates