Docker Isaac Ros Visualisation Rviz problem

Ok in the moment my output is the following but always I see the right image in isaac ros but the reference to the left optical center. Do you have a hint how I can totally disable odom frame and map frame for example. I want to mount the camera to a robot arm and need a hand eye calibration so that I can work in kartesian space. Is it also possible to move the robot now with this informations of the kinematic chain? Do you have there suggestions or interesting descriptions for calibration? I am also interesting in segmentation and ai model development.

My output is in the moment:

header:
stamp:
sec: 1769120782
nanosec: 219822227
frame_id: zed_left_camera_frame_optical
detections:

  • family: tag36h11
    id: 7
    center:
    x: 306.8768310546875
    y: 194.22166442871094
    z: 0.0
    corners:
    • x: 315.0
      y: 199.0
      z: 0.0
    • x: 302.0
      y: 203.0
      z: 0.0
    • x: 298.0
      y: 189.0
      z: 0.0
    • x: 312.0
      y: 185.0
      z: 0.0
      pose:
      header:
      stamp:
      sec: 0
      nanosec: 0
      frame_id: ‘’
      pose:
      pose:
      position:
      x: -0.01473187655210495
      y: 0.03242024406790733
      z: 0.4228404760360718
      orientation:
      x: -0.18981526792049408
      y: -0.392056941986084
      z: 0.8937793374061584
      w: 0.10686426609754562
      covariance:
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0
      • 0.0

I recommend you read our Robot Integration documentation. It refers to a ground robot setup, but the key concepts are the same:

This I already did integrating in urdf. My question is how I can do disable the map and odom frame because on the arm I do not need it. Also the calibration there was not mentioned.

My goal was to test the reference to the base coordinate system from the camera.

Simply set publish_tf:=false in the launch command:

ros2 launch zed_wrapper zed_camera.launch.py camera_model:=<your_camera_model> publish_tf:=false