Checkpoint 4

Last Updated: Nov 20, 2025

The MBot is equipped with a camera that can be used for vision-based tasks, such as identifying AprilTags and detecting obstacles in the environment.

Contents

• Contents
• Computer Vision
  • Task 4.1 Camera Calibration
  • Launch files
  • Video Demo
• Checkpoint Submission

Computer Vision

Task 4.1 Camera Calibration

1. Find a checkerboard in the lab.

   The checkerboard size is defined by the number of intersections, not the number of squares.

   For example, size 8x6 means the board has 8 intersections along one dimension and 6 along the other, corresponding to a grid of 9×7 squares. You’ll need this information for calibration.

2. Run the camera node in the VSCode Terminal:

    ros2 run camera_ros camera_node --ros-args \
    -p orientation:=180 \
    -p width:=640 -p height:=480 \
    -p format:=BGR888
   

3. On NoMachine, run the following command, the GUI will pop up.

    ros2 run camera_calibration cameracalibrator \
    --size 6x8 --square 0.025 --ros-args -r image:=/camera/image_raw -p camera:=/camera
   

   • --size specifies the number of intersections (as noted in Step 1).
   • --square defines the size of each checkerboard square in meters.
   • Adjust the values based on your checkerboard. 0.025 means each square is 25 mm.
4. How to use the calibrator GUI? Refer to the ROS2 Camera Calibration Tutorial
   • Start reading from: Tutorial Steps → Step 5: Start the camera calibration node.
5. Once calibration is complete, move the results to your MBot workspace by running the following commands:

    cp /tmp/calibrationdata.tar.gz ~/mbot_ws/src/mbot_vision
    cd ~/mbot_ws/src/mbot_vision
    tar -xvf calibrationdata.tar.gz -C calibration/
   

   • This will unzip the calibration results to ~/mbot_ws/src/mbot_vision/calibration.
6. Rebuild the packages so the calibration file is accessible system-wide:

    cd ~/mbot_ws
    colcon build
    source install/setup.bash
   

Launch files

To save time on ros2 run all the node individually, you can use the provided launch files.

• To launch camera node only, run the following in the VSCode Terminal:

    ros2 launch mbot_vision camera.launch.py
  

  • This will publish topics /camera/image_raw and /camera/camera_info
  • You can view the camera feed in rqt on NoMachine or Foxglove after launching this node.
• To launch the camera with image rectification using the calibration results (camera node + rectification node), run the following in the VSCode Terminal:

    ros2 launch mbot_vision camera_rectify.launch.py
  

  • You need to do camera calibration first to run the rectification node. It needs the camera info.
  • This will also publish the topic /image_rect
  • You can view the camera feed in both raw and rectified in rqt after launching these 2 nodes.
• To launch full vision pipeline (camera + rectification + AprilTag detection), run the following in the VSCode Terminal:

    ros2 launch mbot_vision apriltag.launch.py
  

  • This will publish AprilTag detection results on /detections.
  • It will also publish the TF from /camera to each Apriltag, so you do not need to decode /detections or convert tag coordinates from the 2D image plane into the 3D world frame yourself.

You can visualize the camera view in rqt on NoMachine desktop:

ros2 run rqt_image_view rqt_image_view

Or on foxglove studio by running the bridge:

ros2 launch foxglove_bridge foxglove_bridge_launch.xml

Video Demo

Checkpoint Submission

No submission is required for this checkpoint.

However, ensure your camera can successfully detect AprilTags, you’ll need this for the competition.