This work is developed from here, which is based on out-dated rosbuild system. I made it support catkin system, and write a more detailed README to make user can build their development environment step by step.

openni2_tracker is a ROS Wrapper for the OpenNI2 and NiTE2 Skeleton Tracker. This is designed as a companion package to the openni2_camera package (found here). Currently, all this node does is publish TF frames of the current tracked user's joint locations.

Note: These instructions only been tested with ASUS Xtion Pro Live in ROS-kinetic, Ubuntu 16.04. You can directly access to the docker image file which support all software development environment within OpenGL visualization at here, or just run the following command to get the docker image:

docker pull bluebirdpp/openni2:latest

1. Install OpenNI2:

   sudo apt install git libusb-1.0-0-dev libudev-dev
   sudo apt install openjdk-8-jdk  # for xenial; openjdk-6-jdk for trusty; if not using other java version.
   sudo apt install freeglut3-dev
   cd  # go home
   mkdir -p src && cd src  # create $HOME/src if it doesn't exist; then, enter it
   git clone https://github.com/occipital/OpenNI2.git  # We used to have a fork off 6857677beee08e264fc5aeecb1adf647a7d616ab with working copy of Xtion Pro Live OpenNI2 driver.
   cd OpenNI2
   make -j$(nproc)  # compile
   sudo ln -s $PWD/Bin/x64-Release/libOpenNI2.so /usr/local/lib/  # $PWD should be /yourPathTo/OpenNI2
   sudo ln -s $PWD/Bin/x64-Release/OpenNI2/ /usr/local/lib/  # $PWD should be /yourPathTo/OpenNI2
   sudo ln -s $PWD/Include /usr/local/include/OpenNI2  # $PWD should be /yourPathTo/OpenNI2
   sudo ldconfig

2. Install ASUS Xtion Pro Live OpenNI driver:

   sudo apt install libopenni-sensor-primesense0

3. Install NiTE2.2:

   cd  # go home
   mkdir -p src && cd src  # create $HOME/src if it doesn't exist; then, enter it
   wget https://sourceforge.net/projects/roboticslab/files/External/nite/NiTE-Linux-x64-2.2.tar.bz2
   tar xvf NiTE-Linux-x64-2.2.tar.bz2
   sudo ln -s $PWD/NiTE-Linux-x64-2.2/Redist/libNiTE2.so /usr/local/lib/  # $PWD should be /yourPathTo/NiTE-Linux-x64-2.2/..
   sudo ln -s $PWD/NiTE-Linux-x64-2.2/Include /usr/local/include/NiTE-Linux-x64-2.2  # $PWD should be /yourPathTo/NiTE-Linux-x64-2.2/..
   sudo ldconfig

   To make sure your NiTE2.2 can run successfully, you need to run some NiTE built-in binary file:

   cd ~/src/NiTE-Linux-x64-2.2/Samples/Bin
   ./UserViewer

   If you got a depth images stream, that means your NiTE work as expected. If you got "no device found" error, you may need to do this:

   sudo ln -s /lib/x86_64-linux-gnu/libudev.so.1.6.4 /lib/x86_64-linux-gnu/libudev.so.0

4. Install orocos_kdl

   cd ~/src
   git clone https://github.com/orocos/orocos_kinematics_dynamics.git
   sudo apt-get -qq update
   sudo apt-get install libeigen3-dev libcppunit-dev
   cd orocos_kinematics_dynamics/orocos_kdl
   mkdir build
   cd build
   cmake -DENABLE_TESTS:BOOL=ON -DCMAKE_CXX_FLAGS:STRING="-Wall" -DCMAKE_BUILD_TYPE=${OROCOS_KDL_BUILD_TYPE} ./..
   # compile and install orocos_kdl
   make -j`nproc`
   sudo make install

5. Clone openni2_tracker to your ROS workspace:

   cd ~/catkin_ws/src
   git clone [email protected]:msr-peng/openni2_tracker.git

6. Configure CMake:

   If you followed Step 1 and Step 2 strictlly, then you needn't do anything. Otherwise, you need to modify CMakeList.txt in openni2_tracker package to make your project can find OpenNI2 and NiTE2.2

7. Make openni2_tracker:

   cd ~/catkin_ws
   catkin_make

8. Set up NiTE2:

   Right now, NiTE requires that any executables point to a training sample directory at .../NiTE-Linux-x64-2.2/Samples/Bin/NiTE2. If you run the NiTE sample code, this works fine because those examples are in that same directory. However, to be able to roslaunch or rosrun openni2_tracker from any current directory, I have created a workaround script setup_nite.bash. This script creates a symbolic link of the NiTE2 directory in your .ros directory (the default working directory for roslaunch / rosrun). You will need to modify this file so that it points to YOUR NiTE2 and .ros locations. By default, you need do this:

   cd ~/catkin_ws/src/openni2_tracker
   ./setup_nite.bash

   I would be pleased if anyone has a better solution to this.

9. Run openni2_tracker:

   roslaunch openni2_tracker tracker.launch

   Then you can visualize the skeleton tracking results on Rviz:

   rosrun rviz rviz -f /tracker_depth_frame

   Finally in Rivz, add TF to visualize the skeleton keypoints coordinates.

   In the lauch file, you can rename both the tracker name and the tracker's relative frame. I have included a static publisher that aligns the tracker frame to the world frame, approximately 1.25m off the floor.

   <!-- openni2_tracker Launch File -->
   <launch>
   
     <arg name="tracker_name" default="tracker" />
     
     <node name="tracker" output="screen" pkg="openni2_tracker" type="tracker" >
       <param name="tf_prefix" value="$(arg tracker_name)" />
       <param name="relative_frame" value="/$(arg tracker_name)_depth_frame" />
     </node>
   
     <!-- TF Static Transforms to World -->
     <node pkg="tf" type="static_transform_publisher" name="world_to_tracker" args=" 0 0 1.25 1.5707 0 1.7707  /world /$(arg tracker_name)_depth_frame 100"/> 
   
   </launch>

   Currently, this node will broadcast TF frames of the joints of any user being tracked by the tracker. The frame names are based on the tracker name, currently /tracker/user_x/joint_name