How Extended kalman filter work in pyro

I visited Pyro ekf example but i could not understand how kalman filter is working in pyro. I don’t see any similar or clear or easy to understanding steps as kalman Filter it self. any comment or source to understand it easier? thanks in advanced.

the tutorial glosses over a lot of the background and is just an example of how to use the library. i’m not too sure what you are asking. to see how the ekf distribution is implemented, you can take a look at the docs. for a more general background on object tracking, if you google “[extended] kalman filter”, you’ll find a lot of great papers and tutorials.

Thanks for your reply. I will check KFDistribution docs as well as other links it could be helpful. But as I told I know how simple or traditional (e)kalman filter work. but the problem is that as an end user programmer and not developer of Pyro. it is not straight forward to implement kalman filter in pyro based on the example and description. as if it is going to be implemented in python or any other languages. in the example it is just some state vector, covariance and noise distribution pass to some pyro function. different distribution like halfcuchy, ncv.process_noise_dist, or dist.MultivariateNormal has been used with out pointing out why they have been used. output plot code is also not available in order to manipulate the parameter to see it’s influences.
The example for a beginner and non developer of Pyro seems too vague. it is not clear how to use the library with those brief explanation. Do you have further recommendation.