search

introlab / rtabmap_ros

RTAB-Map's ROS package.
http://wiki.ros.org/rtabmap_ros
BSD 3-Clause "New" or "Revised" License
1.01k stars 559 forks source link

Is using visual feature-based approaches? #1091

Open kento77 opened 11 months ago

kento77 commented 11 months ago

Sorry for my poor English.

I'm trying to understand how RTABMap works, but could you please help me with the following questions?

1.I am using depth camera realsense455. In this case, is the default setting of RTABMap using a method that uses reprojection error of visual feature-based approaches?

  1. If so, is only 3D pose to be optimized? Or is the 3D point cloud of landmarks optimized in addition to 3D pose?
matlabbe commented 11 months ago
  1. Yes, default parameters should be fine, unless resolution is higher than 720p (parameters are tuned for 480p).
  2. In visual odometry, both pose and 3D features are optimized.
kento77 commented 11 months ago

Thanks for the reply. Is there a paper or something that has the formulas for pose and 3D features optimized.

matlabbe commented 11 months ago

This will give you the overview of the approach https://introlab.3it.usherbrooke.ca/mediawiki-introlab/images/7/7a/Labbe18JFR_preprint.pdf