cartographer icon indicating copy to clipboard operation
cartographer copied to clipboard

Return submap ids in InsertionResult along with pointers

Open ojura opened this issue 6 years ago • 8 comments

Returning just submap pointers in InsertionResult is not very useful for libcartographer users. You can't really tell which submap IDs changed when insertion happened. Therefore, have the pose graph return the assigned submap IDs just like it does for the inserted node ID.

Also, move GetSubmapData to the pose graph interface, so the user can query the pose graph just for these submap IDs (instead of having to use GetAllSubmapData).

ojura avatar Sep 06 '18 05:09 ojura

Ping?

ojura avatar Dec 08 '18 14:12 ojura

@pifon2a since we're on a roll, if I could choose a PR review to get as a Christmas present :), it would be this one. I have encountered multiple times so far a need to know which submaps exactly changed with a recent insertion, and I think this would be a really useful extension of the API.

ojura avatar Dec 17 '18 20:12 ojura

@ojura nice try :). PR review is not really a good Christmas present. I hope you get a normal present instead. Merry Christmas!

pifon2a avatar Dec 17 '18 21:12 pifon2a

Damn :) Merry Christmas and happy new year to y'all as well!

ojura avatar Dec 18 '18 08:12 ojura

Another month has passed, so here's another friendly ping for this PR.

ojura avatar Jan 18 '19 15:01 ojura

@ojura Are you still interested in getting this merged? If so, please let me know and I will review. Also, it seems this PR is doing two completely separate things. Would it be possible to split off the GetSubmapData change to a separate PR? Should be faster to get it merged then. For the other part: can you give some context how it is used, for example the use case you have in mind?

wohe avatar Aug 26 '20 11:08 wohe

Hi @wohe. One application is vertex coloring of scan matched point clouds according to the inserted submap ID (e.g. of the older one from the insertion pair). If you use that golden ratio HSV scheme, I think it’s a nice way to provide false coloring for checking the cloud quality.

Sure, I can split the PR.

ojura avatar Aug 27 '20 09:08 ojura

@wohe here's a picture of coloring by submap ID: image

ojura avatar Aug 27 '20 10:08 ojura