A Jupyter-notebook based article investigating (useful parameter settings for) the Unscented Kalman Filter (UKF).
Everything is explained in the notebook file (UKF.ipynb). Additional python files (*.py) contain some code that I outsourced to keep the main article more human-readable. For backwards-compatibility with reading habits I also provide a pdf, but I strongly advise to try the notebook first. Note, especially, that the figures that I included manually in the notebook are missing in the pdf. The best way to view the notebook online is through Jupyter's notebook viewer service:
http://nbviewer.jupyter.org/github/sbitzer/UKF-exposed/blob/master/UKF.ipynb
Please use Github's issue tracker to post questions and comments.
You can cite this article, for example, as: Bitzer, S. The UKF exposed: How it works, when it works and when it’s better to sample. Zenodo, 2016, doi:10.5281/zenodo.44386