Skip to content

README.md

Latest commit

 

History

History
98 lines (74 loc) · 3.75 KB

README.md

File metadata and controls

98 lines (74 loc) · 3.75 KB
Raw

Foxglove Studio layouts

Web-based replay views for aqua_localization public-data demos. Use the free Foxglove Studio web app — no local install required, no account needed for local-file playback.

Why Foxglove (vs RViz)

  • Runs entirely in the browser; nothing to install.
  • Reads ROS 2 .mcap bags directly via drag-and-drop.
  • Layout is captured as a single JSON file you can commit and share — anyone who drags the same bag onto Foxglove with this layout imported sees the same panels, camera angle, and plots.
  • Decent screenshot / export workflow for README assets.

Available layouts

Layout Demo Bag
aqua_tank_demo.json Tank Dataset short_test — IMU + pressure + DVL fusion vs AprilTag GT recorded via datasets/tank_dataset_demo.md

Recording a results-included bag

Foxglove and rerun.io both play whatever is in the bag — by themselves the source public bags do not contain aqua_localization output topics. Record a "demo" bag that captures both the inputs and our pipeline outputs, so the resulting .mcap is self-contained and reproducible.

One-shot recorder per dataset (each builds on the same pattern: estimator in the background, ros2 bag record writing to mcap, ros2 bag play --clock driving the source bag in the foreground):

ros2 run aqua_localization record_tank_demo.sh
ros2 run aqua_localization record_mbes_demo.sh
ros2 run aqua_localization record_ntnu_demo.sh
ros2 run aqua_localization record_aqualoc_demo.sh

Each script honors environment overrides for source/output paths and replay duration — see the header comment of each aqua_localization/scripts/record_*_demo.sh file.

The Tank demo bag contains ~27 k messages over ~15 s including 5 321 /aqua_imu_loc/odometry samples; the MBES demo bag is ~50 MB across 60 s of multibeam fans plus our pipeline outputs; the NTNU bag is ~21 MB over 90 s; AQUALOC clocks in around 400 MB because of the camera stream.

If you prefer the manual three-terminal flow:

# Source bag (input only).
SRC=aqua_localization/datasets/public/tank_dataset/short_test_ros2
# Output bag (input + estimate + TF).
OUT=aqua_localization/datasets/public/tank_dataset/demo_with_estimate

source install/setup.bash

# Terminal A: estimator.
ros2 run aqua_imu_loc imu_loc_node --ros-args \
  --params-file install/aqua_imu_loc/share/aqua_imu_loc/config/tank_dataset.yaml \
  -p use_sim_time:=true

# Terminal B: recorder.
ros2 bag record -s mcap -o "$OUT" \
  --topics /imu/data /pressure /dvl/twist /apriltag_slam/GT \
           /aqua_imu_loc/odometry /aqua_imu_loc/status \
           /tf /tf_static

# Terminal C: replay.
ros2 bag play "$SRC" --clock

After the source bag finishes, ctrl-C the recorder and estimator.

Loading the layout in Foxglove

  1. Open https://studio.foxglove.dev/ in any modern browser.
  2. Open local file… → select the .mcap from the demo_with_estimate directory above.
  3. Layout menu (top-right) → Import from file… → choose the layout JSON in this directory.
  4. Press play. The 3D panel follows base_link; the green trail is the AprilTag GT, the blue trail is the aqua_imu_loc estimate, the orange arrows are the DVL body-frame twist samples. The right column shows depth (estimate vs GT) and DVL velocity components.

Capturing screenshots / video

Foxglove has no built-in image export. The standard workflow:

  • Single frame: pause at the moment of interest, take an OS screenshot of the browser viewport.
  • Animated GIF / MP4: use OBS or wf-recorder (Wayland) / simplescreenrecorder (X11) on the browser tab; Foxglove's own framerate matches whatever your screen captures.

Save README-bound assets under docs/media/.