Only released in EOL distros:   electric fuerte

Overview

OpenNI is an open-source framework for "natural interaction" - using your hands and body to interact with your digital devices. See OpenNI.org for more information about OpenNI.

Most ROS users will not need to use OpenNI directly. openni_camera already implements a fully-featured ROS camera driver on top of OpenNI. See openni_kinect to get started with your Kinect in ROS.

In Electric and Diamondback, to use OpenNI directly in your packages you would depend on the openni wrapper package. As of Fuerte the openni package is deprecated, and you should link against the system installation of OpenNI as described in the migration guide.

Repositories

Willow Garage maintains the system installs of OpenNI (the framework) and ps-engine (the device driver from PrimeSense) required by ROS. These track the OpenNI and PrimeSense/Sensor projects on GitHub.

We apply some patches to the GitHub sources (e.g. for recognizing Kinects, as well as the ASUS Xtion and PrimeSense's prototype devices). These patches and build scripts are located in our openni/drivers project.

Fuerte migration guide

manifest.xml (depending on OpenNI)

Instead of depending on the openni package, declare the following rosdeps in your manifest.xml:

<rosdep name="openni-dev" />
<rosdep name="ps-engine" />

Now rosdep install MY_AWESOME_PACKAGE will install OpenNI and the ps-engine device driver.

If you provide headers included by code in other packages, and those headers use OpenNI, then you must also export the flags to find OpenNI:

<export>
  <cpp cflags="-I/usr/include/openni" />
</export>

CMakeLists.txt (linking against OpenNI)

The OpenNI headers are installed in /usr/include/openni. The libraries are in the standard search path /usr/lib. In your CMakeLists.txt, add the include directory:

include_directories(/usr/include/openni)

And link your binaries:

rosbuild_add_executable(my_awesome_demo src/my_awesome_demo.cpp)
target_link_libraries(my_awesome_demo OpenNI)

my_code.cpp (using OpenNI in your code)

You can access the OpenNI C++ wrapper in your code with:

   1 #include <XnCppWrapper.h>
   2 

See the OpenNI documentation for more information.