Are ROS 2 teams building internal data pipelines, or just relying on rosbag2?
I’m curious how serious robotics teams are handling real-world data collection in ROS 2.
For small experiments, recording topics with rosbag2 and writing custom scripts seems fine.
But once a team starts collecting data for robot learning, teleoperation, debugging, or fleet analytics, the workflow can get messy quickly.
For example:
-camera streams
-joint states
-force / torque data
-operator commands
-robot state topics
-failure cases
-teleop sessions
-rosbag2 replay data
The hard part seems to be less about recording data, and more about turning it into something clean, searchable, synchronized, and useful later.
For teams using ROS 2 in production or serious research:
Are you mostly using rosbag2 + custom scripts, or has your team built a more structured internal data pipeline for collection, storage, filtering, replay, and training?