Kalman tutorial - terrytaylorbonn/auxdrone GitHub Wiki

24.1027

The goal of this tutorial is to create the material a reader needs to get started hands-on with Kalman filters ASAP. With a minimum of text and a maximum of useful content. Any mistakes I have made or misconceptions I have about the tech details I will figure out with time (I am doing this with no assistance and no previous Kalman filter experience).

I have created (WIP) the following 4 docs (gdrive) that explain (based on my understanding and with hands-on examples) the very basics of Kalman filters:

  • (FIRST DRAFT) 1_Kalman_demo_1_(Becker)_.docx. Hands-on basic demo part 1 (constant velocity, short time interval; based on Becker). Uses the demo to help you grasp the concepts.
  • (TODO) 2_Kalman_concepts_(Becker)_.docx. Concepts (based on Becker) focusing more on how the algorithms work. Much of this I still struggle with, not sure of the content.
  • (TODO) 3_Kalman_demo_2_(Becker)_.docx. Hands-on basic demo part 2 (with acceleration, longer time interval; based on Becker).
  • (WIP) 4_Kalman_python_notebooks_overview_(Labbe)_.docx. Overview of python notebook demos (excellent material, extensive) (based on Labbe). This worked well on my PC. Its more about using the API.

Note:

  • The examples used in doc 1 are based on a demo from Carlos Becker and a book from Alex Becker. The demo does not match the book, and I would have done the example differently. But they are the examples I chose to start with.
  • If I was creating this material for a customer, the latest working version would always be available internally (perhaps on a wiki like this one) for constant internal feedback/review. After many revisions, rewrites, and expert approvals, I would then publish the first public version with professional quality text and graphics.

(FIRST DRAFT) 1_Kalman_demo_1_(Becker)_.docx

This doc is based on a step-by-step demo from CppMonk (Carlos Becker) (Git, Video) which discusses

  • Ch 1 "Gist" of the Kalman filter (my understanding)
  • Ch 2 Running the Becker demo code
  • Ch 3 Code (and plot / console output) analysis
  • Ch 4 Tuning the Kalman filter

image

(TODO) 2_Kalman_concepts_(Becker)_.docx

Will be based on kalmanfilter.net (also from Carlos Becker?) but

  • leaves out the many many confusing equation derivations
  • references the results from the doc 1 demo
  • will try to make an intuitive analysis of the equations

image

(TODO) 3_Kalman_demo_2_(Becker)_.docx

Adds more hands-on tests.

image

(WIP) 4_Kalman_python_notebooks_overview_(Labbe)_.docx

Will be based on a great Python notebook from R. Labbe.

My old doc from 2 months ago: 20.1_32_kalman_rlabbe_v06_24.0902.docx

image

image