Add an auto-camera calibration script

[amalnanavati]

Description

This PR adds an auto-camera calibration script. Through command-line interaction, it interactively guides the user to place a charucoboard in front of the robot (the parameters for the charucoboard should already be specified in config/calibrate_camera.yaml). Then, the robot moves the arm in a star around its hardcoded starting configuration, varying roll and pitch as it goes, to gather samples. It then runs eye-in-hand camera calibration using those samples, using the config-specified method (I empirically found that Andreff worked best).

Note that when computing the translation and rotation error, we first get the predicted charucoboard pose in robot frame for all of the samples. We then compute the translation/rotation distances between every pair of samples, and take the median of those distances.

This procedure should be re-run every time the eye-in-hand system is modified, e.g., unscrewed/rescrewed, some 3D printed components were replaced, etc.

Testing

• Pull this branch, re-build the workspace.
• Modify the parameters in config/calibrate_camera.yaml to your charucoboard as necessary.
• Test online camera calibration:
  • Run the camera calibration code: python3 src/ada_ros2/ada_calibrate_camera/calibrate_camera_start.py
  • Follow the on-screen instructions to gather samples calibrating the camera.
  • Once the procedure is done, copy-paste the translation and rotation error below.

Rotation error: 0.016380652571024775
Translation error: 0.004512129207414376
R_cam2gripper: [ 0.00238821 -0.12084312  0.99259871  0.01179148]
R_cam2gripper: [ 3.1187588  -0.00759098 -0.24220932]
T_cam2gripper: [[ 0.03562513]
 [ 0.13481861]
 [-0.05481037]]

- [x] Terminate camera calibration code: `python3 src/ada_ros2/ada_calibrate_camera/calibrate_camera_start.py -c`

• Test offline camera calibration:
  • Change config/calibrate_camera.yaml to run offline, point it to the new data directory that was created above, and run all the camera calibration methods.
  • Run ros2 run ada_calibrate_camera calibrate_camera --ros-args --params-file <path/to/ada_ros2>/ada_calibrate_camera/config/calibrate_camera.yaml
  • Paste the translation and rotation errors across methods below (same as above)
• Test the feeding demo with the new calibration:
  • Launch the feeding demo: python3 src/ada_feeding/start.py
  • Test the acquisition of 3 bites, verify the camera is centered.

[amalnanavati]

TODO: Change to ~/Workspace to align with lovelace