loop closure rejected: not enough feature in images (old=0, new=179, min=20)

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

loop closure rejected: not enough feature in images (old=0, new=179, min=20)

bobi
This post was updated on .
Hello,

I already post this problem on github, but posting on the official forum could help.

I keep getting this warning :
Rtabmap.cpp:3036::process() Rejected loop closure 104 -> 1110: Not enough features in images (old=0, new=181, min=20).
Values are changing but the problem is still the same. Old is always 0 even if the mapping is working.
I tried adding or replacing some args in my launchfile, but nothing.

I'm using a RGB camera and a lidar.

None of my loop closure aren't rejected.

[Edit1 : I detect loop closure when i put RGBD/Enabled false. But then the map on Rviz and on Rtab-map is not shown, nothing is displayed]


Second thing: If I’m in localization mode and detect loop closures, will my drone re-localize itself using those loop closures, even if it doesn’t know its position in the mapped room?

Thanks for your help
Reply | Threaded
Open this post in threaded view
|

Re: loop closure rejected: not enough feature in images (old=0, new=179, min=20)

matlabbe
Administrator

See answer on same issue on github: https://github.com/introlab/rtabmap_ros/issues/1259

When setting RGBD/Enabled to false, rtabmap will only "localize" in term of images, no metric poses will be computed/updated. That is a mode most people will never use, unless you are evaluating only the appearance-based loop closure detection approach of rtabmap.

bobi wrote
Second thing: If I’m in localization mode and detect loop closures, will my drone re-localize itself using those loop closures, even if it doesn’t know its position in the mapped room?
Yes, it will then know its position in the room after re-localizing in it.