Hybrid-Astar-Planner

This project hosts trajectory planner based on Hybrid A* for unstructured environment

Documentation

This project consists of several ROS packages, which are:

• Part 1: trajectory planner
  • mpros_planner: node for trajectory planner
  • mpros_algorithm: path planner with Hybrid A* algorithm
  • mpros_msgs: trajectory messages for communication between planner and controller
  • mpros_navigation_map: preprocessing block, generating different map representations
  • mpros_rs_path: Reeds-Shepp path generator
  • mpros_smoother: path optimizer with Ceres
  • mpros_speed_profile: speed profile generator according to S-Curve
  • mpros_utils: utils functions for planner

Dependencies

• ROS Noetic
  see ROS Noetic Installation
• Eigen3, version >= 3.4
  • Download source code from Eigen Releases
  • Extract compressed file
  • Find the path to local eigen like /usr/include/eigen3
  • Install new Eigen lib by replacing old version

  cd eigen-3.4.0
  cmake .. -DCMAKE_INSTALL_PREFIX=/usr # to replace eigen 3 in /usr/include/eigen3

  • check installation by

  std::cout << EIGEN_WORLD_VERSION << "."
            << EIGEN_MAJOR_VERSION << "."
            << EIGEN_MINOR_VERSION << std::endl;

• OMPL
  • see OMPL Installation, recommend installation with ROS as shown below
  • Installation with ROS as follows

  sudo apt-get update
  sudo apt-get install ros-noetic-ompl

• CERES
  • follow the Ceres Installation

Usage - To be tested due to removed simulation environment

• Clone all mpros package to your workspace which is shown below as catkin_ws:

  cd ~/catkin_ws/src
  # clone this repo to local
  git clone [email protected]:Runningchauncey/Hybrid-Astar-Planner.git
  # make the workspace
  cd ~/catkin_ws
  source /opt/ros/noetic/setup.bash
  catkin_make
  source devel/setup.bash
  # launch the simulation environment
  roslaunch mpros_planner vehicle_test.launch case:=0
  # run planner node in another terminal
  source devel/setup.bash
  rosrun mpros_planner main_planner

• Set up initial pose and goal pose in Rviz with 2D Pose Estimate and 2D Nav Goal by press the button, click a position and drag the arrow to set orientation

Configure

• The packages mpros_algorithm, mpros_navigation_map, mpros_rs_path, mpros_smoother, mpros_speed_profile are configurable by load parameter file in .yaml format
• Parameter explanation see README in each package
• Configuration files under mpros_planner/config
• Load a configuration file while simulation environment on by

  rosparam load /path/to/file

Test cases

All test cases can be load by change the case number when launch simulation environment.

• 0: reverse parking
• 1: parallel parking
• 2: Arena part
• 3: Arena whole map
• 4: slalom map
• 5: parking lot

TODO

• Custom simulation environment
• Dubins Curve generation