🤖 AI Summary
Rerun has added experimental support for MCAP, the open-source multimodal log format that became the default ROS2 bag format in May 2023, allowing users to visualize MCAP robot logs in the Rerun viewer and SDK without changing workflows or formats. You can drag-and-drop MCAP files into the viewer, run them with the CLI (pip install rerun-sdk; rerun recording.mcap), or convert MCAP to Rerun’s .rrd (rerun mcap convert ...) including per-layer conversions. This is an early release, but it opens up easier inspection and visualization of existing robotics datasets and aligns Rerun with growing ROS2 adoption.
Technically, Rerun maps MCAP’s topic/message model into its entity-component system via a layered architecture: log_time and publish_time from MCAP are preserved as two distinct timelines; files are ingested into multiple layers (raw, ros2msg, protobuf, schema/stats/recording_info) that can be queried independently. The ros2msg layer provides semantic visualizations for common ROS2 types (sensor_msgs, geometry_msgs, std_msgs, builtin_interfaces)—images -> Image, point clouds -> Points3D, IMU -> SeriesLines—while the protobuf layer uses reflection to decode schemas into Arrow types and structured components (queryable but not automatically visualized). Important limitations remain (custom messages, tf transforms, ROS1, and other types not yet mapped), and Rerun requests community feedback to prioritize further message support.
Loading comments...
login to comment
loading comments...
no comments yet