*Jul*

The previous illustration is in this link This post is mainly talking about the error fixed in the MM estimator. The biggest error of the MM estimator in derivation and code is that: The error is: In the matrices, I used the scalar 1 to replace identity matrix I. For example, using (1-K) to re...

*Jun*

The previous handwriting derivation is in this link

*Jun*

Big picture The deficiency of the current derivation is marked red color, which would be extended in the next week.

*Apr*

Those slides are presenting the evolution of input estimation based on Kalman Filter. I picked up 14 papers as the representatives of those evolution. Next week I would implement the 2d Multi-model and test to see how effective the new MM algorithm perform better than the native one.

*Mar*

The source code of this project is https://git.uclalemur.com/zhz03/DARPA_CREATE_Code/-/tree/master/

*Mar*

This notes are quick notes after I have tested all Create code in practice. It contains the terms which easy to make people puzzled and somethings left to do. I marked this note in case to forget before I continue to finish the left parts. Important terms in code: binary_1d: u (event) is binary...

*Feb*

Next task: The urgent tests need to be carried out is to find the real pose estimation result only by landmark points without uncertainty evaluate the influence of uncertainty to find the pose estimation effect based on probability map of features After that we can then to test the EM alg...