search

introlab / rtabmap_ros

RTAB-Map's ROS package.
http://wiki.ros.org/rtabmap_ros
BSD 3-Clause "New" or "Revised" License
922 stars 549 forks source link

guidance for rendering 3D maps from /rtabmap/mapData topic in a web browser #1037

Closed 360wcui closed 9 months ago

360wcui commented 9 months ago

Is there any example of rendering 3D maps from /rtabmap/mapData in a browser?

I am able to extract the messages using react-ros and roslibjs (attached below). Any advice on visualizing them in JavaScript would be greatly appreciated.

Screenshot from 2023-09-24 08-47-24

matlabbe commented 9 months ago

You would have to reconstruct the point clouds from the depth image and calibration. In rtabmapviz and rviz's mapcloud plugin, we create a point cloud for each node in the map. On loop closures, we only need to update the point cloud's pose.

360wcui commented 9 months ago

Is /voxel_cloud the consolidated point clouds from the each node? If not, what is the difference? Thank you.

MRSHIMMER commented 2 months ago

@360wcui Hi, I want to visualize the 3D reconstructed model in real time in a web application. Can I directly use the data in ‘/rtabmap/mapData’? But it's not clear to me what the data in this topic means.

matlabbe commented 2 months ago

You may use /rtabmap/cloud_map directly (PointCloud2). To avoid streaming the whole cloud at each update, then you would need to subscribe to mapData topic. It is more complicated to handle thought, but it has everything you need (rgb image, depth image, calibration, optimized poses of all nodes in the graph).

For the voxel_cloud that @360wcui was refering, it is the cloud created from latest camera data (at same framerate than camera topics) mainly used in the examples for visualization purpose or for obstacle detection, it is not an ouput of rtabmap node.

MRSHIMMER commented 2 months ago

You may use /rtabmap/cloud_map directly (PointCloud2). To avoid streaming the whole cloud at each update, then you would need to subscribe to mapData topic. It is more complicated to handle thought, but it has everything you need (rgb image, depth image, calibration, optimized poses of all nodes in the graph).

For the voxel_cloud that @360wcui was refering, it is the cloud created from latest camera data (at same framerate than camera topics) mainly used in the examples for visualization purpose or for obstacle detection, it is not an ouput of rtabmap node.

Thank you very much for your answer, and I also want to thank you for your contribution to this project. To avoid unnecessary expenses, I will try using data from the 'mapGraph' topic. I am currently using my phone and it is not convenient to access information. My initial idea is to refer to rviz's mapcloud plugin. Thank you again.