Foxglove

Physical AI Industry Night is back! We're again inviting the SF robotics community to join us for drinks, a panel discussion, and networking on November 17th.

Our last Industry Night in May brought together 300 robotics developers and had a waitlist of 1,200+ registrants.

This event will fill up fast, so RSVP soon. All details on the Luma page -> hubs.li/Q04C2Zxs0

The full panel for Physical AI Industry Night is here!

Join Foxglove and the SF robotics community for drinks, networking, and a discussion on triaging and debugging robotics data at scale. Hear how teams test and QA robot behavior across mountains of real-world data, and how their workflows evolve as their fleets grow.

November 17 @ 5:30PM | San Francisco

Thank you to our sponsors, @nebiusai and @anyscalecompute, for supporting the night!

Space is limited. Link in the comments.

New in Foxglove: An XY plot with Sync with other plots now follows the time window of the plots it is synced with. Points inside the window stay at full opacity and the rest dim, so zooming a timestamp plot shows you which part of a path happened in that interval. Every point stays on screen, and the path doesn't rescale as the window moves.

Libby Lin developed a system for autonomous fleet mapping that displays data from multiple vehicles in a global OctoMap.

The results are interesting: Foxglove is used to stitch maps from multiple vehicles, and we stand by this approach as long as you can define a common frame of reference across all vehicles in your fleet! We look forward to seeing where Libby takes this project next!

Learn more: hubs.li/Q04B_B7j0

Credit: @Libby Lin

Join our Discord to stay up to date on our community news: hubs.li/Q04B_B890

One SO-101, no leader arm, and a full record-and-replay setup in ROS 2!

Aditya Kamath is back with the next part of his SO-101 series. This time the arm learns to repeat a motion.

How it works:
- Record a motion by hand-guiding the limp arm with torque off, or by teleoperating it with IK or joint sliders
- Joint states and TF go straight into an MCAP bag through rosbag2_py
- Replay runs through the same path as teleop, so every pose gets the joint limits and self-collision checks
- Loop a recording once, N times, or until you pause or e-stop

Then the fun part. He replayed the same motion on the real arm and in mock mode and lined up both MCAP files in Foxglove Comparison Mode. The median error was 0.2 to 2.4° per joint.

You can open both recordings yourself and scrub through the comparison.

Read the full post: hubs.li/Q04Cc0vP0

#ros #lerobot