This repo reproduces the code for the paper PhaRaO: Direct Radar Odometry using Phase Correlation. Although the original paper is based on iSAM, this release integrate the code with gtSAM.

1. ROS The code has been tested using ROS Noetic

2. GTSAM Use develop branch.

3. FFTW3 (fftw-3.3.10 version is recommended for ubuntu 20.04)

   We need normal installation followed by float option installed. We need both FFTW3Config.cmake file and FFTW3fConfig.cmake. So first compile without float option.

   $ cmake .
   $ ./configure --enable-shared --enable-threads --enable-float
   $ make
   $ sudo make install

   Then, do the same thing with the float option on.

   $ cmake .
   $ ./configure --enable-shared --enable-threads --enable-float
   $ make
   $ sudo make install

   If FFTW3LibraryDepend.cmake file-related error occurred, try the below lines. Check this issue

   $ sudo cp FFTW3LibraryDepends.cmake /usr/local/lib/cmake/fftw3/

4. Download Radar Data samples (ex. MulRan)

1. Generate your own catkin workspace & Go to your src folder

   mkdir ~/catkin_ws
   cd ~/catkin_ws
   mkdir src
   cd src
   

2. Download the source file for PhaRaO

   git clone https://github.com/hyesu-jang/PhaRaO_gtsam.git
   

3. Compile the source file

   cd ~/catkin_ws
   catkin_make
   

4. Launch File Player Module(MulRan for example)

   roslaunch file_player file_player.launch
   

5. Run the PhaRaO roslaunch file

   roslaunch pharao_gtsam radar_odom.launch
   

  5-1. Or run by PhaRaO node and rviz

 rosrun pharao_gtsam pharao_gtsam_node
 rviz -d ~/catkin_ws/src/PhaRaO_gtsam/rviz/odom.rviz

  5-2. Saving result trajectory

Enable save flag "True" <param name="save_results_flag" value="false" />

If your bag file fails to generate odometry, revise the parameter in the file radar_odom.launch

1. Cost for generating odometry factor

2. Cost for keyframe factor

3. Scale Factor

4. Thresholds for linear and angular velocity

Hyesu Jang (dortz at snu dot ac dot kr) Yeong Sang Park (pys0728k at gmail dot com)