Michel Hidalgo

+1000 to this.

Yeah, I'm aware. I've been postponing a resolution because, as you say, > It's ok to ignore the rolling build because it's not guaranteed but sooner or later we will...

> When initialized in an unknown region, the initial particle cloud collapses **because unknown pixels in the map are treated as obstacles by the beam sensor**. Both nav2 and beluga...

> I think we can remove the "bug" label from this, as everything seems to be working as expected in its design. Fair, if we take the beam model design...

I like the idea.

> As a side note, they [don't seem to use the cubic formula to aggregate probabilities](https://github.com/ros-navigation/navigation2/blob/85fd06f20413e348059205a84ecb50765fc8f0c7/nav2_amcl/src/sensors/laser/likelihood_field_model_prob.cpp#L232) in this model, so we should look into that as well. They use log...

> I wonder if, instead of implementing it, we should evaluate whether this actually makes a difference in performance. Is there empirical or bibliographic support for this feature? Does people...

All in all, in general and from a process perspective, I'd like to have a standard procedure for validating improvement hypothesis such as this. Landing https://github.com/Ekumen-OS/lambkin/pull/93 and #415 will get...

> There's neither other than being a long standing feature in Nav2. FTR I unearthed https://github.com/ros-planning/navigation/pull/238. Those numbers they promised never saw the light of day, and yet it was...

> we know the sum of cubes works because it's the default and therefore we can infer people have been using it for years with no complains I'm with you...