Reasoning behind "key" attribute in AMCL k-d tree

edit

Hi everyone,

I am working on a project that extends AMCL and I need to understand its design. I got stuck on the particle filter, namely the clustering, k-d tree and the "key" field. The main part of code in question is here: https://github.com/ros-planning/navig...

Could anyone please explain to me how is value for the "key" determined? Which dimension is used during node creation, how is the k-d tree built and how is decided, which nodes belong to the same cluster? If these is any documentation for this please point me to it.

Thanks in advance :-)