Table Calibration

The table calibration is partly automatic table position estimation using Xtion sensor. The calibration method uses OpenCV to find a corner in a small neighbourhood of a table corners defined in the URDF model in the lower resolution. This position is further refined in the full resolution. This is don for all four corners of a table. Depth image is then used to fit a table plane.

  • Start robot (in separate terminal)
    $ roslaunch clopema_launch start_robot.launch
  • Start Xtion (in separate terminal)
    $ roslaunch clopema_launch xtionN.launch
  • Start table calibration node (in separate terminal)
    $ rosrun clopema_calibration -x N -t 2
  • I windows should pop up where one can see and image with rough positions (red) and refined position (green) and their distance, move the t robot so that all four corners (the real and green) are visible and move the table corners close to the green corners so the detection algorithm can catch the right spot.
  • When done press Ctr + c in the terminal window to store the calibration file.
  • In order to remake the model run the following command and restart robot to see the calibration result.

Note: during the calibration the table corners are published in TF and table position is published as a Marker in /table_calibration topic.