Find solution
How to apply kalman filter on rotation matrix?

What's a switching Kalman Filter? How is it different from a regular Kalman Filter and HMMs?

  • Searching for a nice intro summary of this topic, assuming the reader has a basic understanding of HMMs and Kalman Filters already.

  • Answer:

    My out-of-nowhere guess is that a switching KF is a form of multi-modal Kalman filtering where instead of linearly combining each model based on their respective variances, maybe you just select the model with the lowest variance.  (One could really imagine a number of different ways to combine the output from multiple KFs.) If you want another example of using a KF in situations with non-Gaussian noise (e.g. outliers), I wrote a paper that wrestled with that.  I created a (very simple) "bad state detector" that had various recovery modes if the error residuals got too high.  http://scholar.google.com/scholar?cluster=11075302979446480557&hl=en&as_sdt=2000

Adam Smith at Quora Visit the source

Was this solution helpful to you?

Other answers

It's been a while.  Here is what I can understand from this paper: http://www.cs.berkeley.edu/~murphyk/Papers/skf.ps.gz Kalman filters assume linear systems with Gaussian noise, so they have obvious real-world limitations.  HMM attempts to solve this by discretizing hidden state variables.  This has the limitations of poor inference capability and requiring large initial data sets to train the system. Switching Kalman filters form a number of separate linear models and then switch between them or linearly combine them

Eugene Bialczak

Related Q & A:

Just Added Q & A:

Find solution

For every problem there is a solution! Proved by Solucija.

  • Got an issue and looking for advice?

  • Ask Solucija to search every corner of the Web for help.

  • Get workable solutions and helpful tips in a moment.

Just ask Solucija about an issue you face and immediately get a list of ready solutions, answers and tips from other Internet users. We always provide the most suitable and complete answer to your question at the top, along with a few good alternatives below.