ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

openni_tracker displaying pointcloud and skeleton alignment?

asked 2013-02-20 13:50:46 -0600

jys gravatar image

updated 2016-10-24 09:01:28 -0600

ngrennan gravatar image

I have openni_tracker working. However, I am not sure how I can visualize both point cloud from Kinect as well as the skeleton TF data.

I can visualize Kinect point cloud by setting fixed frame to /camera_link

And I can visualize TF skeleton data by setting fixed frame to /openni_depth_frame.

When I visualize TF data, there is no connection between point cloud related TF and skelecton related TF data. Am I doing something wrong?

How could I align these together?

image description image description

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2013-02-21 11:17:03 -0600

updated 2013-03-05 08:27:51 -0600

The easiest solution would be to launch the openni_launch with the camera argument set. See the documentation at the bottom of this page. So the following command

roslaunch openni_launch openni.launch camera:=openni

will set the first word of all of the frames associated with your Kinect to be "openni" instead of "camera". This should merge your tf tree.

edit flag offensive delete link more


@jarvisschultz would you recommend openni_tracker for people detection?? Or do you know ay good stack that would acommplish people detection? Thanks

ctguell gravatar image ctguell  ( 2013-09-12 09:59:43 -0600 )edit

@ctguell if all you wanted to do was detect people, you could use openni_tracker, but it might be a bit overkill. If you still wanted to use a depth camera, maybe you could adapt openni_tracker a bit e.g. see the `User_NewUser` callback in the openni_tracker source.

jarvisschultz gravatar image jarvisschultz  ( 2013-09-12 17:00:53 -0600 )edit

As of 1.7, PCL now has pretty good people detection built in, if that's really what you want (as opposed to human pose estimation). Have a look at this tutorial:

Neil Traft gravatar image Neil Traft  ( 2014-08-13 16:48:06 -0600 )edit

Question Tools

1 follower


Asked: 2013-02-20 13:50:46 -0600

Seen: 1,087 times

Last updated: Mar 05 '13