Posted by
GGYU on
URL: http://official-rtab-map-forum.206.s1.nabble.com/RTABMAP-Multi-RGBD-Camera-Error-tp11093p11107.html
β [Question]
I have successfully rebuilt RTAB-Map with OpenGV, and previous errors were resolved with noticeable performance improvements.
However, Iβm now encountering new issues:
π¨ [Current Issues]
Velocity commands via teleop occasionally get dropped β possibly due to high computational load.
RViz drops messages with Message Filter dropping message errors, indicating that the message queue is full.
Misalignment between odometry (local map) and the generated global map β causes confusion in localization and visualization.
The RTAB-Map GUI (rtabmapviz) does not appear or launch β which makes it difficult to debug graph and loop closure status.
β
[What I Want to Achieve]
I want to perform 2D grid-based mapping using LIDAR scan data (as in turtlebot3_sim_rgbd_scan demo),
while utilizing 3 RGB-D cameras solely for:
Detecting ArUco markers (for loop closure or localization refinement),
Extracting and matching visual features for visual loop closure optimization.
π [My Question]
Is this hybrid approach possible with RTAB-Map?
If so, what is the recommended architecture or launch configuration to make this work?
In particular, Iβd appreciate guidance on:
How to integrate scan-based mapping with RGB-D inputs only contributing to loop closures,
How to properly configure RTAB-Map to separate visual odometry and LIDAR-based mapping,
Best practices for handling high data load from multiple RGB-D cameras.
Thank you in advance for your help.
μ€ν¬λ¦°μΊμ€νΈ_04-23-2025_07-37-29_PM.mp4