bulletsim_source

Code for tracking deformable objects by stereo cameras

Tang, Te, Yongxiang Fan, Hsien-Chung Lin, and Masayoshi Tomizuka. "State estimation for deformable objects by point registration and dynamic simulation." In Intelligent Robots and Systems (IROS), 2017 IEEE/RSJ International Conference on, pp. 2427-2433. IEEE, 2017.

Tang, Te, Changhao Wang, and Masayoshi Tomizuka. "A framework for manipulating deformable linear objects by coherent point drift." IEEE Robotics and Automation Letters 3, no. 4 (2018): 3426-3433.

I. Dependency Installation

1. Install Ubuntu 14.04 LTS

Reference: http://howtoubuntu.org/how-to-install-ubuntu-14-04-trusty-tahr

2. Install ROS-Indigo

2.1 Installation

Reference: http://wiki.ros.org/indigo/Installation/Ubuntu

sudo sh -c 'echo "deb http://packages.ros.org/ros/ubuntu $(lsb_release -sc) main" > /etc/apt/sources.list.d/ros-latest.list'
sudo apt-key adv --keyserver hkp://ha.pool.sks-keyservers.net:80 --recv-key 421C365BD9FF1F717815A3895523BAEEB01FA116
sudo apt-get update
sudo apt-get install ros-indigo-desktop-full
sudo rosdep init
rosdep update
echo "source /opt/ros/indigo/setup.bash" >> ~/.bashrc
source ~/.bashrc
sudo apt-get install python-rosinstall

2.2 Create Catkin workspace

Reference: http://wiki.ros.org/ROS/Tutorials/InstallingandConfiguringROSEnvironment

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws/
catkin_make
echo "source ~/catkin_ws/devel/setup.bash" >> ~/.bashrc
source ~/.bashrc

3. update opencv to include the opencv-nonfree module

sudo add-apt-repository --yes ppa:xqms/opencv-nonfree
sudo apt-get update 
sudo apt-get install libopencv-nonfree-dev

4. Driver for Kinect V1

sudo apt-get install libfreenect-dev
sudo apt-get install ros-indigo-freenect-launch

Test Kinect V1 connection by

roslaunch freenect_launch freenect.launch

5. Driver for Kinect V2

Your computer needs to have USB3.0

Install the latest versoin of Nivida GPU driver. Restart computer after GPU driver installation:

sudo apt-get install nvidia-3 (press tab and install the one with the largest version number, such as 340, 375)

Use ocl-icd-libopencl1 to replace nvidia-libopencl1

sudo apt-get install ocl-icd-opencl-dev

Install libfreenect2

cd ~
git clone https://github.com/OpenKinect/libfreenect2.git
cd libfreenect2
cd depends; ./download_debs_trusty.sh
sudo apt-get install build-essential cmake pkg-config
sudo dpkg -i debs/libusb*deb
sudo apt-get install libturbojpeg libjpeg-turbo8-dev
sudo dpkg -i debs/libglfw3*deb; sudo apt-get install -f
cd ..
mkdir build && cd build
cmake .. -DCMAKE_INSTALL_PREFIX=$HOME/freenect2 -DENABLE_CXX11=ON
make
make install
sudo cp ../platform/linux/udev/90-kinect2.rules /etc/udev/rules.d/

Plug in Kinect V2 to USB3.0, run the test program for libfreenect2

cd bin
./Protonect gl
./Protonect cl
./Protonect cpu

Install IAI Kinect2 to bridge the libfreenect2 and ROS

cd ~/catkin_ws/src/ 
git clone https://github.com/code-iai/iai_kinect2.git 
cd iai_kinect2 
rosdep install -r --from-paths . 
cd ~/catkin_ws 
catkin_make -DCMAKE_BUILD_TYPE="Release"

Test whether the kinect data is published as ROS Topic

roslaunch kinect2_bridge kinect2_bridge.launch
rostopic list

6. Other dependency for bulletsim

sudo apt-get install libopenscenegraph-dev python-networkx python-scipy

II. Install bulletsim

1. Download code and set up the environment variables

cd ~
mkdir DeformableTracking && cd DeformableTracking
git clone git@github.com:thomastangucb/bulletsim_source.git
mkdir bulletsim_build
echo 'export BULLETSIM_SOURCE_DIR=~/DeformableTracking/bulletsim_source' >> ~/.bashrc
echo 'export BULLETSIM_BUILD_DIR=~/DeformableTracking/bulletsim_build' >> ~/.bashrc
source ~/.bashrc

2. Compile ROS packages "bulletsim_msgs" and "bulletsim_python" for customized ROS messages

2.1 Copy the "bulletsim_msgs" and "bulletsim_python" folders to Catkin workspace

cp -R $BULLETSIM_SOURCE_DIR/src/bulletsim_msgs ~/catkin_ws/src
cp -R $BULLETSIM_SOURCE_DIR/src/bulletsim_python ~/catkin_ws/src

2.2. Catkin_make the ROS package and create a project for Eclipse

cd ~/catkin_ws
catkin_make --force-cmake -G"Eclipse CDT4 - Unix Makefiles"

Check whether the customized messages and srvices ("bulletsim_msg/...") are recognized by ROS

rosmsg list 
rossrv list

2.3. Add path to system PYTHONPATH

To run initialization_service.py (in bulletsim_python) in terminal, add python path to .bashrc.

echo 'export BULLETSIM_PYTHON_DIR=~/catkin_ws/src/bulletsim_python/src' >> ~/.bashrc
echo 'export PYTHONPATH=$PYTHONPATH:$BULLETSIM_PYTHON_DIR' >> ~/.bashrc
source ~/.bashrc

3. Compile "fgt" and "cpd" packages for rope tracking

3.1 Copy the "fgt" and "cpd" folders to outside

cp -R $BULLETSIM_SOURCE_DIR/lib/fgt ~/DeformableTracking
cp -R $BULLETSIM_SOURCE_DIR/lib/cpd ~/DeformableTracking

3.2 Compile package "fgt"

mkdir ~/DeformableTracking/fgt/build && cd ~/DeformableTracking/fgt/build
cmake ..  -DCMAKE_BUILD_TYPE=Release
make && sudo make install
make test

3.3 Compile package "cpd"

mkdir ~/DeformableTracking/cpd/build && cd ~/DeformableTracking/cpd/build
cmake ..  -DCMAKE_BUILD_TYPE=Release -DWITH_FGT=ON
make && sudo make install
make test

4. Create two Eclipse projects (Release and Debug) to folder "bulletsim_build"

cd $BULLETSIM_SOURCE_DIR
./make_eclipse_project.sh

5. Import the Release and Debug projects to Eclipse, then compile

reference: http://www.cnblogs.com/cv-pr/p/4871546.html

If compiling failed [Fatal error: gl/glut.h: no such file or directory], then install freeglut3 package

sudo apt-get install freeglut3-dev

III. Test Tracking

1. Use Kinect for online experiment

Make a link between every binary in the build folder and a new ros package named "binary_symlinks", such that roslaunch can find these binaries

roscd
roscreate-pkg binary_symlinks && cd binary_symlinks
mkdir bin && cd bin
for node in $BULLETSIM_BUILD_DIR/release/bin/* ; do ln -s $node ; done

Run roslaunch to start three steps together (1)kinect data logging (2)camera calibration (3)point cloud preprocessing:

roslaunch bulletsim_msgs kinect2.launch calibrationType:=0

Here, calibrationType means whether you need to calibrate the kinect or you have already calibrated it. Everytime you move the Kinect, set calibrationType to 1, put a chessboard on the table and launch the file for once and then set it back to 0 for the test.

Run initialization_service.py

python $BULLETSIM_PYTHON_DIR/tracking_initialization/scripts/initialization_service.py

Run tracker_node_CPD

$BULLETSIM_BUILD_DIR/release/bin/tracker_node_CPD

(a) kinect1.launch

For one Kinect V1 use, includes connect with Kinect 1, calibration and downsample the topic (change topic name) and launch preprocessor_color_node.

(b) kinect2.launch

For one Kinect V2 use, includes connect with Kinect 2, calibrate and downsample the topic (change topic name) and launch preprocessor_color_node.

(c) kinect12.launch

For one Kinect V1 and one Kinect V2 use, includes connect with Kinects, calibrate and downsample the topic (change topic name) and launch two preprocessor_segmentation_node. If you plan to use two kinects, an inputTopic config is needed to change, which is on the config_tracking.cpp, add the second kinect name, such as /kinect2 after

std::vector<std::string> TrackingConfig::cameraTopics = boost::assign::list_of("/kinect1");

2. Use Kinect for online experiment (step by step without using roslaunch)

https://docs.google.com/a/berkeley.edu/document/d/1rV39njt_-qcb-sSsAd2HKh0qMhbHoNoIG9QPp373nAg/edit?usp=sharing

3. Use recorded data for offline test

Download bagfile from our shared Google Drive folder, put them into ~/DeformableTracking/bulletsim_source/data/bagfiles/ Start ROS kernel

roscore

Run initialization_service.py

python $BULLETSIM_PYTHON_DIR/tracking_initialization/scripts/initialization_service.py

Run tracker_node_CPD

$BULLETSIM_BUILD_DIR/release/bin/tracker_node_CPD

Play the recorded point cloud

rosbag play $BULLETSIM_SOURCE_DIR/data/bagfiles/testrope.bag

If everything is installed correctly, when the rosbag is palyed, a window will immediately pop up and shows a virtual object that follows the motion of point cloud.