hybrid_path_planning

This repository contains ROS2 Humble packages implementing hybrid path planning algorithms for vehicle that can move both forward and backward, including reeds_shepp and dubins motion

📥 Install dependencies

sudo apt-get install libeigen3-dev
sudo apt install ros-$ROS_DISTRO-joint-state-publisher-gui
sudo apt install ros-$ROS_DISTRO-xacro
sudo apt install ros-$ROS_DISTRO-vision-opencv
sudo apt install ros-$ROS_DISTRO-grid-map-ros
sudo apt install ros-$ROS_DISTRO-grid-map-cv
sudo apt install ros-$ROS_DISTRO-cv-bridge
sudo apt install libopencv-dev

📋 Distribution

The Reeds-Shepp Hybrid A algorithm: incorporates both forward and reverse motion to generate optimal paths for non-holonomic vehicles. It computes a set of possible steering angles and directions and simulates vehicle movement to evaluate potential trajectories, checking for collisions and boundary conditions in a grid map. Costs are calculated based on movement direction (forward or reverse), steering angle, and changes in direction, with higher penalties for reversing and sharp turns. Additionally, the algorithm includes a heuristic function for holonomic motion planning, where obstacles are considered when calculating costs, and it uses the Reeds-Shepp curve to optimize the path when no valid solution is found in a straightforward trajectory.

The Dubins Hybrid A algorithm: generates paths using only forward motion, leveraging Dubins curves for optimal path planning with non-holonomic constraints. The algorithm simulates vehicle motion based on a set of forward-only steering commands, checking for collisions and computing the cost of each path based on steering angles and direction changes. A Dubins path is calculated between the current and goal nodes, and the cost is based on the length of the path and steering complexity. Additionally, a holonomic heuristic is used to account for obstacles and to optimize the path when standard Dubins paths are not feasible.

⚙️ Configuration

This project includes various configuration options for different nodes, allowing for customization of vehicle parameters, map handling, and path planning settings. These configurations are located in the global_launcher at the config file. Below is an explanation of the configuration parameters for each node.

Dubins Path Planning Node:

• maxSteerAngle: The maximum steering angle of the vehicle in radians. This value controls how sharp the vehicle can turn. (Default: 0.6)
• wheelBase: The distance between the front and rear axles of the vehicle. This affects the vehicle's turning radius. (Default: 2.75)
• axleToFront: The distance from the front axle to the front of the vehicle. This is important for calculating the vehicle's geometry. (Default: 1.1)
• axleToBack: The distance from the rear axle to the back of the vehicle. This is used for vehicle geometry calculation. (Default: 1.2)
• width: The width of the vehicle. This parameter affects collision checking and how the vehicle fits in the environment. (Default: 1.9)
• pathLength: The maximum length of the planned path in meters. (Default: 4.0)
• step_car: The step size for moving the vehicle forward in the simulation. A smaller value results in finer control but more computation. (Default: 1.0)
• grid_map_topic: The ROS topic that provides the grid map used for path planning. (Default: "/grid_map_inflated")

Reeds-Shepp Path Planning Node:

• maxSteerAngle: Same as above, the maximum steering angle in radians. (Default: 0.6)
• wheelBase: The distance between the front and rear axles. (Default: 2.75)
• axleToFront: The distance from the front axle to the front of the vehicle. (Default: 1.1)
• axleToBack: The distance from the rear axle to the back of the vehicle. (Default: 1.2)
• width: The width of the vehicle for collision checking. (Default: 1.9)
• pathLength: The maximum length of the planned path in meters. (Default: 4.0)
• step_car: Step size for moving the vehicle in the simulation. (Default: 1.0)
• grid_map_topic: The ROS topic for the grid map used in path planning. (Default: "/grid_map_inflated")

Grid Map Publisher:

• resolution: The resolution of the grid map, specifying how much area each cell represents in meters. A lower value provides higher detail. (Default: 0.2 or 1.0)
• grid_map_topic: The topic on which the grid map will be published. (Default: "/grid_map")
• image_dir: The directory path where the image file is stored. This image is converted into a grid map. (Default: "/home/genis/personal/ros2_ws/src/hybrid_path_planning/grid_map_publisher")
• image_file: The name of the image file used to create the grid map. (Default: "gridmap.png")

** Map Inflate Node**:

• inflation_radius: The radius in meters around obstacles that should be inflated to account for the vehicle’s size. (Default: 0.2)
• grid_map_topic_IN: The input topic where the original grid map is published. (Default: "/grid_map")
• grid_map_topic_OUT: The output topic where the inflated grid map will be published. (Default: "/grid_map_inflated")

📍 Build

 colcon build --packages-select dubins_hybrid_star
 colcon build --packages-select reeds_shepp_hybrid_astar
 colcon build --packages-select map_inflate
 colcon build --packages-select grid_map_publisher
 colcon build --packages-select global_launcher

▶️ Run

• Launch Reeds-Shepp Hybrid Path Planner:

ros2 launch global_launcher reeds_shepp_hybrid.launch.py

• Launch Dubins Hybrid Path Planner:

ros2 launch global_launcher dubins_hybrid.launch.py

🔴 Notes:

• ❗ Important: If you are publishing a map or using a specific grid map for planning, make sure to update the paths in the configuration file. This includes the image_dir and image_file paths for the grid_map_publisher. ❗
• Make sure your ROS2 environment is properly sourced before launching the files (source ~/ros2_ws/install/setup.bash).
• You can modify the parameters in the config/global_launcher.yaml file to suit your vehicle and environment.
• The planners require a grid map to operate, so ensure that a valid grid map is published to the specified topics (/grid_map_inflated).