Underwater Data Tools

This repository contains the script for convertting ROS bag to RAW data format of Tank Dataset.

Prerequisites

• Docker
• Docker Compose

Setup

The project uses Docker to provide a consistent ROS Noetic environment with all necessary dependencies.

1. Build the Docker Image

docker compose build

This will:

• Set up a ROS Noetic environment
• Clone and build the waterlinked_a50_ros_driver package
• Install all Python dependencies (OpenCV, NumPy, etc.)
• Configure user permissions to match your host system

2. Start the Container

docker compose up -d

Using bag2raw.py

The bag2raw.py script extracts data from ROS bag files and converts them into images (PNG) and CSV files.

What it extracts:

• Images:

  • Left camera images (/camera/left/image_dehazed/compressed) → IMG_L/
  • Right camera images (/camera/right/image_dehazed/compressed) → IMG_R/

• CSV Files:

  • gt.csv - Ground truth pose from AprilTag SLAM (/apriltag_slam/GT)
  • gt_full.csv - Full ground truth pose (/apriltag_slam/GT_full)
  • aqua_slam_pose.csv - AQUA SLAM pose estimates (/aqua_slam/pose)
  • depth.csv - Depth measurements (/depth/data)
  • imu.csv - IMU data (angular velocity, linear acceleration) (/imu/data)
  • dvl.csv - DVL velocity data (/dvl/data)

Usage

Place your .bag files under data folder, then run:

docker exec -it underwater_data_tools bash
python3 ./bag2raw.py ./data

If your bag files are outside the workspace, you need to mount them first in compose.yaml file.

Output Structure

For each .bag file processed, a directory with the same name will be created containing:

your_bagfile/
├── IMG_L/
│   ├── 1234567890123456789.png
│   ├── 1234567890234567890.png
│   └── ...
├── IMG_R/
│   ├── 1234567890123456789.png
│   ├── 1234567890234567890.png
│   └── ...
├── gt.csv
├── gt_full.csv
├── aqua_slam_pose.csv
├── depth.csv
├── imu.csv
└── dvl.csv

Note: Image filenames are timestamps in nanoseconds (format: secs * 1e9 + nsecs).

CSV File Formats

gt.csv, gt_full.csv, aqua_slam_pose.csv

timestamp[ns], pos_x[m], pos_y[m], pos_z[m], orient_qx, orient_qy, orient_qz, orient_qw

depth.csv

timestamp[ns], pos_z[m]

imu.csv

timestamp[ns], ang_vel_x[rad/s], ang_vel_y[rad/s], ang_vel_z[rad/s], lin_acc_x[m/s^2], lin_acc_y[m/s^2], lin_acc_z[m/s^2]

dvl.csv

timestamp[ns], vel_x[m/s], vel_y[m/s], vel_z[m/s], beam1_vel[m/s], beam2_vel[m/s], beam3_vel[m/s], beam4_vel[m/s]

Troubleshooting

Permission Issues

If you encounter permission issues, ensure the .env file exists with your user IDs:

echo "USER_ID=$(id -u)" > .env
echo "GROUP_ID=$(id -g)" >> .env

Then rebuild:

docker compose build --no-cache

ROS Messages Not Found

If you see errors about missing message types (especially DVL messages), ensure the catkin workspace was built correctly:

docker compose exec ros_noetic bash -c "source /opt/ros/noetic/setup.bash && source /catkin_ws/devel/setup.bash && rospack list | grep dvl"

Out of Memory

Processing large bag files may require significant memory. You can limit the processing or increase Docker's memory allocation in Docker Desktop settings.

Dependencies

• ROS Noetic
• Python 3
• OpenCV
• NumPy
• rosbag
• waterlinked_a50_ros_driver (DVL message definitions)

All dependencies are automatically installed in the Docker container.

LICENSE

This project is licensed under the GPL-3.0 License. See the LICENSE file for details.