Feature/4 dof for point to plane minimizer #378
Conversation
Minor fixes in the code to follow github reviewers suggestions.
|
Can one of the admins verify this patch? |
|
add to whitelist |
|
|
||
| - OrientNormalsDataPointsFilter: | ||
| towardCenter: 1 | ||
|
|
There was a problem hiding this comment.
Are the extra lines between the reference data points filters intended?
There was a problem hiding this comment.
I like it better, easier to distinguish between the filters.
|
I just approved this PR, it's a really cool contribution. Just out of curiosity: did you observe any improvement in the CPU usage or accuracy of ICP with respect to full DOF minimization? |
|
There should not be a significant speedup, the difference is solving a 4x4 matrix instead of 6x6 one. I don't expect much speedup, the heavy lifting is done elsewhere in ICP... |
|
Ok. All good. |
|
@pomerlef Sorry to bother. Are you OK with the changes introduced by this MR? |
|
All good, go ahead |
|
@simonpierredeschenes Simon Pierre, could you press the button please? Thx |
This PR adds the possibility to optimize only translation and yaw angle instead of the full 6DOF pose. A unit test has been also added. Since this feature is a part of research in progress, there are no publications yet. The code has been successfully used during the DARPA SubT challenge, so I consider it quite well tested.