Ask Your Question

pgorczak's profile - activity

2018-06-26 04:06:30 -0500 received badge  Nice Answer (source)
2018-06-25 10:15:30 -0500 answered a question Consequence of setup.py

Correct! As long as you're using the devel space, catkin generates the file you mentioned via a cmake macro. When it is

2018-06-25 09:42:18 -0500 answered a question Piping catkin_make stderr to less while preserving color

The less command does not receive the color control chars because it is not a terminal. Not sure if you can pipe output

2014-11-19 17:59:55 -0500 answered a question Building PCL on OSX 10.9

I could get the latest version to install on OS X Yosemite with brew install pcl --HEAD

Some more discussion in a related question.

2014-11-19 17:56:41 -0500 answered a question Can't install PCL with Homebrew on OSX 10.9 during ROS Hydro installation

On OS X Yosemite, the normal brew install pcl failed as well for me. Adding --HEAD did the trick again.

2014-04-23 23:48:03 -0500 received badge  Favorite Question (source)
2013-08-27 17:42:35 -0500 received badge  Enlightened (source)
2013-08-27 17:42:35 -0500 received badge  Guru (source)
2013-02-25 04:06:21 -0500 commented answer OpenNI/NITE incompatible in Fuerte/Groovy on Precise

Having the same issues here. "niReg -l" still lists libXnVFeatures.so and libXnVHandGenerator.so as "compiled with OpenNI 1.0.0.22" when installing ros-groovy-openni-tracker (so these come with the system dependency libopenni-nite-dev). For a listing of the output see Kel Guerin's post.

2013-02-25 04:06:21 -0500 received badge  Commentator
2013-01-17 10:16:43 -0500 received badge  Nice Answer (source)
2013-01-14 04:05:37 -0500 commented answer Setting up ROS Groovy - can't launch roscore

As a side note, after uninstalling rospkg, rosinstall and rosdep using pip, I had to reinstall the system packages python-rospkg, python-rosinstall and python-rosdep (with apt-get install --reinstall).

2012-12-14 01:35:20 -0500 received badge  Great Question (source)
2012-09-28 10:30:42 -0500 received badge  Good Question (source)
2012-09-04 13:52:45 -0500 received badge  Nice Question (source)
2012-09-04 01:45:51 -0500 received badge  Famous Question (source)
2012-09-03 01:24:18 -0500 commented question OpenNI/NITE incompatible in Fuerte/Groovy on Precise

Are you sure the camera is running? Try to have openni_launch openni.launch up at the time you run openni_tracker. If the camera is working properly, you should be able to see an image with image_view. If you walk around in front of it, openni_tracker should give output then.

2012-08-31 23:00:57 -0500 received badge  Notable Question (source)
2012-08-31 07:46:36 -0500 received badge  Great Answer (source)
2012-08-31 07:06:20 -0500 commented answer openni_tracker Timeout Error in Fuerte (Ubuntu 12.04 LTS)

Hi andyw, I was able to find a workaround for the "oparation invalid" issue which you can find in my updated post (the one you linked to).

2012-08-31 06:27:53 -0500 received badge  Student (source)
2012-08-31 06:27:10 -0500 received badge  Popular Question (source)
2012-08-31 05:50:11 -0500 edited question OpenNI/NITE incompatible in Fuerte/Groovy on Precise

When trying to use openni_tracker with Kinect (Xbox-version) on Ubuntu 12.04 with Fuerte. I was getting the following error message:

[ERROR]: Find user generator failed: This operation is invalid!

I got to know more about what's happening in the backgound now. So the device is queried for a "User generator" which for NITE is XnVSkeletonGenerator. With niReg -l, you can list the available libraries and the OpenNI-nodes they enable. What looks a bit off is, that the OpenNI version is 1.5.2.23 and everything has that version except the libraries relevant for user tracking. Those are versioned as 1.3.0.17 and apparently compiled against OpenNI 1.0.0.22.

After installing the latest unstable "OpenNI Compliant Middleware Binaries" for Ubuntu 12.04 from openni.org, I have several library versions as shown below and the user_tracker works flawlessly. To make it work, don't remove nite-dev (now replaced with libopenni-nite-dev) but run the uninstall.sh from the downloaded archive before the install.sh so you won't break the dependency but replace the libraries.

I created a bug report for that issue so the above workaround hopefully gets obsolete.

Edit: As you can see from the comments (and I got it as well), this issue is also relevant in Groovy. The "nite-dev" dependency got replaced with "libopenni-nite-dev" and the basic OpenNI dependencies are now "libopenni-sensor-primesense-dev" and "libopenni-dev".

niReg -l output for the Fuerte/Precise openni-dev and nite-dev packages:

(...)
OpenNI version is 1.5.2.23
(...)
/usr/lib/libXnVFeatures.so (compiled with OpenNI 1.0.0.22):
    Scene: PrimeSense/XnVSceneAnalyzer/1.3.0.17
    User: PrimeSense/XnVSkeletonGenerator/1.3.0.17

/usr/lib/libXnVHandGenerator.so (compiled with OpenNI 1.0.0.22):
    Gesture: PrimeSense/XnVGestureGenrator/1.3.0.17
    Hands: PrimeSense/XnVHandTracker/1.3.0.17

niReg -l output for the Fuerte/Precise openni-dev and after applying workaround:

(...)
OpenNI version is 1.5.2.23
(...)    
/usr/lib/libXnVFeatures_1_3_0.so (compiled with OpenNI 1.0.0.22):
    Scene: PrimeSense/XnVSceneAnalyzer/1.3.0.17
    User: PrimeSense/XnVSkeletonGenerator/1.3.0.17

/usr/lib/libXnVFeatures_1_3_1.so (compiled with OpenNI 1.2.0.8):
    Scene: PrimeSense/XnVSceneAnalyzer/1.3.1.8
    User: PrimeSense/XnVSkeletonGenerator/1.3.1.8

/usr/lib/libXnVFeatures_1_4_1.so (compiled with OpenNI 1.3.2.3):
    Scene: PrimeSense/XnVSceneAnalyzer/1.4.1.2
    User: PrimeSense/XnVSkeletonGenerator/1.4.1.2

/usr/lib/libXnVFeatures_1_4_2.so (compiled with OpenNI 1.3.4.6):
    Scene: PrimeSense/XnVSceneAnalyzer/1.4.2.5
    User: PrimeSense/XnVSkeletonGenerator/1.4.2.5

/usr/lib/libXnVFeatures_1_5_2.so (compiled with OpenNI 1.5.4.0):
    Scene: PrimeSense/XnVSceneAnalyzer/1.5.2.21
    User: PrimeSense/XnVSkeletonGenerator/1.5.2.21

/usr/lib/libXnVHandGenerator_1_3_0.so (compiled with OpenNI 1.0.0.22):
    Gesture: PrimeSense/XnVGestureGenrator/1.3.0.17
    Hands: PrimeSense/XnVHandTracker/1.3.0.17

/usr/lib/libXnVHandGenerator_1_3_1.so (compiled with OpenNI 1.2.0.8):
    Gesture: PrimeSense/XnVGestureGenrator/1.3.1.8
    Hands: PrimeSense/XnVHandTracker/1.3.1.8

/usr/lib/libXnVHandGenerator_1_4_1.so (compiled ...
(more)
2012-08-31 05:46:20 -0500 received badge  Good Answer (source)
2012-08-31 05:45:57 -0500 received badge  Nice Answer (source)
2012-08-31 05:05:54 -0500 answered a question What would be helpful for newbies to know before getting started?

First I'll try to elaborate a bit on the already mentioned points:

  1. Basic Linux and especially Ubuntu
    • Getting used to using the windows and workspaces relevant shortcuts (in Unity, hold SUPER/WINDOWS)
    • Knowing what apt-get does and how it works
    • Knowing how to set up ssh connections using keys
  2. Using the terminal
    • Getting familiar with TAB completion and history
    • Know the basic shortcuts CTRL+C, CTRL+ALT+T, CTRL+SHIFT+T, CTRL+(+|-|0)

What also helped me:

  1. Having a basic knowledge of threads and their hazards (especially relevant for C++)
  2. Getting familiar with version control systems and their underlying concepts
    • git, subversion and mercurial are all used in ROS repositories
    • Focus on one and use it for your projects (git works best for me)
2012-08-30 23:22:07 -0500 edited answer openni_launch + fuerte + ubuntu 12.04: No image received

The first exception is not harmful as you can see here.

When I am using roslaunch openni_launch openni.launch, I also get messages like:

[ERROR] [1346317018.377804268]: Tried to advertise a service that is already advertised in this node [/camera/depth_registered/image_rect_raw/compressedDepth/set_parameters]
[ERROR] [1346317018.384694527]: Tried to advertise a service that is already advertised in this node [/camera/depth_registered/image_rect_raw/compressed/set_parameters]
[ERROR] [1346317018.413763705]: Tried to advertise a service that is already advertised in this node [/camera/depth_registered/image_rect_raw/theora/set_parameters]
(...)
[ WARN] [1346317020.499849507]: Camera calibration file (...)/rgb_A00364A06244047A.yaml not found.
[ WARN] [1346317020.500131130]: Using default parameters for RGB camera calibration.
[ WARN] [1346317020.500289570]: Camera calibration file (...)/depth_A00364A06244047A.yaml not found.
[ WARN] [1346317020.500426152]: Using default parameters for IR camera calibration.

Still, everything works just fine (image_view and rviz) so those shouldn't be the issue.

I think running only image_view like in your first edit and still getting the freeze pretty much rules out a performance issue--at least concerning rviz. I think (I'm not sure though) the point cloud only gets computed when you subscribe to the points topic which would mean there is pretty much nothing going on computationally when you just display an image with image_view.

You could also try a different Kinect device or make sure your Kinect works e.g. on Windows with the OpenNI examples to make sure there is no problem with the device itself.

And my last question would be: What is your version of ROS and your operating system (version)? I'm using ubuntu 12.04 with fuerte and it worked with the OpenNI-packages I got with sudo apt-get install ros-fuerte-openni-camera.

You could try the driver from openni_camera_deprecated just to see if it gives you different output or some hints on what's going on.

Kinect launch file using openni_camera/OpenNINodelet (deprecated version of the driver):

<launch> 
    <node pkg="nodelet" type="nodelet" name="openni_manager" output="screen" respawn="true" args="manager"/>
    <node pkg="nodelet" type="nodelet" name="openni_launch" args="load openni_camera/OpenNINodelet openni_manager" respawn="true">
        <!--<param name="rgb_frame_id" value="camera_rgb_optical_frame" />-->
        <!--<param name="depth_frame_id" value="camera_depth_optical_frame" />-->
        <param name="depth_registration" value="true" />
        <param name="image_mode" value="VGA_30Hz" />
        <param name="depth_mode" value="VGA_30Hz" />
        <param name="debayering" value="EdgeAwareWeighted" />
        <param name="depth_time_offset" value="-0.055" /> <!-- Taken from TurtleBot's kinect.launch -->
        <param name="image_time_offset" value="0" />
    </node>
</launch>
2012-08-30 23:07:48 -0500 commented answer OpenNI/NITE incompatible in Fuerte/Groovy on Precise

I've also run into that ticket but I think there is some confusion between the "timeout" error and the "operation is invalid" one. Everyone getting the latter one is actually getting past the part where the timout error might occur. But still, I hope someone dealt with it already...