JuliaRobotics / Caesar.jl

Robust robotic localization and mapping, together with NavAbility(TM). Reach out to info@wherewhen.ai for help.
https://www.wherewhen.ai
MIT License
185 stars 31 forks source link

Get basic CJL.ScatterAlignPose3 functional on standardized code #869

Open dehann opened 2 years ago

dehann commented 2 years ago

TAC

get SAP3 functional as simple point to point alignment.

Notes

dehann commented 2 years ago

Image

dehann commented 2 years ago

Wip:

dehann commented 2 years ago

Common rigid transform function for _PCL and ScatterAlign, working torwards Pose3:

dehann commented 2 years ago

Further work for ScatterAlignPose3:

dehann commented 2 years ago

Just adding reference alignment solution from completeness:

dehann commented 2 years ago

To speed things up (and after finding), I'm consolidating the simpleICP.jl implementation. This is the defacto standard. One trick not included in ScatterAlign is to setup the cost function against path normals around each point.

Here is the first piece:

and here is what the code does with the two pointclouds from above:

Image

dehann commented 1 year ago

good chunk of this is probably already done with recent work. So the remaining TAC is that ScatterAlign runs through and has a test, but okay if the numerics are not robust.

dehann commented 1 year ago

I recently also added an option for alignments on either

dehann commented 11 months ago

reminder to consolidate versions across both Caesar and PyCaesar functionality