Ask Your Question

ar_kinect not finding published point cloud data

asked 2011-05-17 15:13:37 -0500

Billy gravatar image

updated 2016-10-24 08:33:43 -0500

ngrennan gravatar image


I'm trying to use ar_kinect ( ) to report pose information of a fiducial, visual marker.

I'm taking the following steps to run ar_kinect:

  1. Open terminal and run: roscore
  2. Open terminal and run the following to start Kinect: roslaunch openni_camera openni_node.launch
  3. Open terminal and run the following so Kinect will begin reporting information on /camera/rgb/camera_info (which is needed by ar_kinect): rostopic hz /camera/rgb/image_mono
  4. Open terminal and run: roslaunch ar_kinect ar_kinect.launch

After doing so, everything seems to start up correctly until it tries to subscribe to image and point cloud topics, as follows:


SIZE = 640, 480 Distortion factor = 0.000000 0.000000 -0.000000 1.000000 525.00000 0.00000 319.50000 0.00000 0.00000 525.00000 239.50000 0.00000 0.00000 0.00000 1.00000 0.00000

header: seq: 0 stamp: 0.000000000 frame_id: points[]: 0 width: 0 height: 0 sensor_origin_: 0 0 0 sensor_orientation_: 0 0 0 1 is_dense: 1

-1.205960 -0.903880 2.478000 [ar_kinect-1] process has died [pid 21130, exit code -11]. log files: /home/lebowski/.ros/log/c3fd81c4-80f9-11e0-835e-001de075d661/ar_kinect-1*.log all processes on machine have died, roslaunch will exit shutting down processing monitor... ... shutting down processing monitor complete done

Interestingly, it looks like it's picking up an empty point cloud based on the dump info shown above. But if I open a terminal and run the following, I see plenty of Kinect data being published on both topics:

rostopic echo /camera/rgb/points rostopic echo /camera/rgb/image_color

Any thoughts?

Thank you very much! Billy McCafferty

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2011-05-17 17:04:40 -0500

fergs gravatar image

updated 2011-05-23 10:23:05 -0500

OK, I've finally had time to sit down and work on this, and this issue should be fixed. In addition to testing on a real Kinect, I spent a bit of time doing some major speed up work. The newly committed version (r263) now does the following:

  • ignores empty clouds
  • subscribes only to the cloud -- uses pcl functions to create the image, which will fix issues with synchronization
  • no longer computes normals (which was slow), we now use PCL functions to estimate the transformation based on the corner points (which also appears to be more accurate than just computing the normal of the center).

Please let me know if you encounter any issues.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools


Asked: 2011-05-17 15:13:37 -0500

Seen: 405 times

Last updated: May 23 '11