The quick answer
Your rosbag already has point clouds
If the file contains sensor_msgs/PointCloud2, the raw points are already there. Local utilities can export them one frame at a time. That is useful, but it is not the same as a map.
Rosbag to point cloud
Turn a rosbag into a point cloud map
Engineers often search for "rosbag to point cloud" when they are really trying to solve a bigger problem than file conversion. The raw recording already contains point clouds. The harder part is turning the whole recording into one usable map.
The real job
A map needs SLAM and sometimes GNSS
A mapped point cloud needs pose estimation across the whole run. If you want geographic coordinates too, you also need a georeferencing step after SLAM.
When "rosbag to point cloud" means "rosbag to map"
The symptoms are pretty obvious once you see them
You want one output from the whole drive or field run, not one file per timestamp.
You need a reproducible browser workflow that other people on the team can review.
You want GNSS-backed alignment instead of only a locally consistent SLAM result.
What LidarFlow changes
It closes the gap between a recording and a mapped artifact
Bring in `.mcap` or ROS1 `.bag` from the browser.
Check LiDAR, GNSS, IMU, and TF before you start the run.
Run GLIM-based LiDAR SLAM and export the mapped point cloud artifacts.
Preview and download the result without handing the run off to another tool.
When local tools are still enough
- You are debugging one topic and only need raw frames.
- You already have a SLAM stack and only need a helper utility.
- You are running a one-off experiment and do not need a shared review surface.
When the browser workflow is the simpler answer
- You want a map back and not more glue code.
- You want the run settings and artifacts logged in one place.
- You want a teammate to review the result without recreating your environment.
Related guides