Poor Man's Explanation of Kalman Filtering [pdf] | Hacker News Reader

Poor Man's Explanation of Kalman Filtering [pdf]

edge.rit.edu

59 pointskqr211 comments

Wow, this is really well done. Everyone should go read this now!

Kalman filtering is one of these techniques that's used everywhere and yet most engineers/developers don't even know it exists (let alone the math behind it, which is fairly straightforward undergrad level stuff.)

Awesome! I worked on the inertial navigation system for Insight Racing's entry into the 2nd DARPA Grand Challenge as my senior design project in college. We implemented a Kalman filter, but I'm not sure we got all the bugs out as it didn't work very well. I sure wish I had known about this little doc at the time!

Actually, I should probably clarify: _our senior design team_ couldn't get the inertial navigation system working very well. I'm pretty sure they took what we worked on and fixed it as they finished somewhere in the middle of the pack. It was easily the coolest project I've ever worked on.

Dan Simon's 2001 article is just as good, if not better IMO. http://web.cecs.pdx.edu/~greenwd/kalman_overview.pdf

"Taygeta Scientific Incorporated, has acquired the rights to republish the report." http://www.taygeta.com/kalman_book.html

Has anyone bought a copy? I assume it is typeset rather than being a very difficult to read fifth generation photocopy scanned to PDF.

Offtopic: Why was kjhgfghjk's joke (this comment's sibling) killed?

It's funny, since the scan of the article is very 'noisy' and a Kalman filter is a possible solution to help clean it up (probably not the best, but still ...)

I would say that in more than half the cases where I see a dead comment, there is nothing obviously very wrong with the comment. Sometimes it's dead because the user who posted it is auto-dead-ified -- everything s/he posts is killed -- but in most of the (few) such cases where I've looked back over the user's comment history, I haven't found anything very awful that they got put in that state for.

Tip: highlight the nasty text: the PDF was done with OCR, so the text is magically cleaned up using "select".

Here is an open source Kalman filter (written for my iPhone app) by a very smart friend of mine: http://github.com/lacker/ikalman

It's in C.

If you just want the equations, try:

http://www.ece.cmu.edu/~byronyu/papers/derive_ks.pdf

And for the EKF:

http://www-npl.stanford.edu/~byronyu/papers/derive_eks.pdf

Welch and Bishop also have an excellent introduction: http://www.cs.unc.edu/~welch/media/pdf/kalman_intro.pdf

Reply on news.ycombinator.com