[image_cb_detector_action_old] doesn't exist for camera_pose_calibration and it will not allow me to use the calibrator. Is there a known work-around? I know there's a but I know there's a [image_cb_detector_action] but I'm unable to make it work. I'm trying to get 3 Kinects to work in Hydro and publish a transform between them in Rviz.
Originally posted by Athoesen on ROS Answers with karma: 429 on 2014-01-08
Post score: 2
camera_pose_calibration has been hopelessly broken for a long time. Due to the official version not working, there was a git hub fork that fixed some of the bugs: http://answers.ros.org/question/69655/what-happened-to-the-stack-camera_pose-in-groovy/#76681
However, I have had no success with the fork either and had to resort to Matlab's camera calibration toolbox. It's strange that there's no "upper level" push to fix such an important tool, but there clearly isn't.
Originally posted by Arrakis with karma: 163 on 2014-01-28
This answer was ACCEPTED on the original site
Post score: 1
Comment by Arrakis on 2014-01-29:
If you don't get the cv_bridge errors with the github version, this additional code is needed if you want to calibrate the Kinect IR camera
http://answers.ros.org/question/55524/asus-extrinsic-calibration-failing-on-ir-image/
Comment by Athoesen on 2014-01-30:
Will this work in ROS Hydro though?
Comment by Arrakis on 2014-02-05:
I run groovy and with a lot of patience and debugging I got it to work. I have not tried Hydro. The last updates to the Git hub code are over a year ago, so I would not be surprised if it didn't work. However, this code is the only available ROS tool to calibrate the extrinsic matrix.
Comment by Athoesen on 2014-02-06:
Yes, which I find incredibly frustrating. As someone who has been trying to run this on Hydro only to find out it's hopelessly busted, I'm basically starting from scratch. Would you suggest going the Groovy route or the Matlab route? I need to have SOMETHING to show in 3 weeks...
Comment by Arrakis on 2014-02-06:
http://pastebin.com/rxgw6LbK
I quickly created a diff of my changes to capture_monitor.py from the Git hub link to remove the 8UC3 bug. This is not a general fix. It's designed for specifically for my Kinect setup. However, maybe this and, if using Kinect, the intensity changing node, can help