Kalman filtering framework

Object framework for filtering using Kalman filter, EKF, or UKF.

Description

Object-based framework for performing Kalman filtering for discrete time systems or continuous-discrete hybrid systems. Includes code for the classical Kalman filter for linear systems, the extended Kalman filter (EKF), and the more recent unscented Kalman filter (UKF). Both linear and non-linear noise in the system and observation are permitted.

Three examples are included in the demos folder.

• MATLAB release: MATLAB 7.8 (R2009a)

Other Files

KalMat/BaseContinuousSystemModel.m,
KalMat/BaseDiscreteSystemModel.m,
KalMat/BaseObserverModel.m,
KalMat/BaseSystemModel.m,
KalMat/BLUEFilter.m,
KalMat/demos/DoubleWell.m,
KalMat/demos/HeatConduction.m,
KalMat/demos/Reentry.m,
KalMat/EKFContinuousSystemModel.m,
KalMat/EKFObserverModel.m,
KalMat/EKFSystemModel.m,
KalMat/ekfTransform.m,
KalMat/LinearContinuousSystemModel.m,
KalMat/LinearObserverModel.m,
KalMat/LinearSystemModel.m,
KalMat/sdchol.m,
KalMat/symmetricSigmaPoints.m,
KalMat/UnscentedContinuousSystemModel.m,
KalMat/UnscentedObserverModel.m,
KalMat/UnscentedSystemModel.m,
KalMat/unscentedTransform.m,
license.txt

[Kenz]

Interesting idea, but the lack of documentation kills it. Perhaps if the examples had been better chosen, for instance to include multidimensional problems, it might have been clearer. As it were, when I tried a simple spring-mass system I got lots of "matrix dimensions must agree" errors, and, being unfamiliar with Matlab's `classdef` features, couldn't really sort out what was going wrong.

[Kenz]

(Excuse me, I wasn't clear. I meant for the "continuous-discrete" functionality, which is where this kind of framework could come in truly handy)