dwl2021 / dynamic-astar Goto Github PK

This repository provides a framework for dynamic Astar (also known as dynamic A*, kinodynamic Astar, knodynamic A*) path searching in a 3D point cloud map.

Dynamic Astar is a common path search algorithm, but there is a lack of a unified framework. Although there are many related repositories on GitHub, very few of them worked and met my expectations.

Based on a great algorithm, Fast-Planner, I extracted the path search method and made it compatible with the point cloud map.

More development details can be found in the ## Development section.

Step 1: Download the dependencies

sudo apt update
sudo apt install ros-$ROS_DISTRO-octomap
sudo apt install ros-$ROS_DISTRO-octomap-server
sudo apt install ros-$ROS_DISTRO-pcl-ros

Step 2: Initialize the repo and compile

git clone https://github.com/Dwl2021/Dynamic-Astar.git
cd Dynamic-Astar
catkin_make

Step 3: Run the program

./run.sh

Modify in src/plan_manage/launch/example.launch

Eigen::Vector3d start(0, 0, 1);
Eigen::Vector3d start_vel(0, 0, 0);
Eigen::Vector3d start_acc(0, 0, 0);
Eigen::Vector3d goal(30, 0, 1);
Eigen::Vector3d goal_vel(-2, 0, 0);

Modify in src/plan_manage/launch/kastar.xml

Modify in src/map_pcl/pcd and if you want to make your own map, more details can be found in src/map_pcl/pcd/map_generation.

If you want to change your own point cloud, please modify the frame to world and the topic to /global_map. Possibly, you might find the path crossing through the point cloud; there may be several reasons:

1. The point cloud is too sparse.
2. The max velocity is too large.
3. The max_tau is too large, which depends on the step time.

Or you can modify the map_util.h file. Search for the expand_size variables and set them to 1, 2, or 3.

If you want to adjust the size of map, please pay attention to the origin point in map_util.h

origin_d_[0] = -map_size(0) / 2;
origin_d_[1] = -map_size(1) / 2;
origin_d_[2] = 0;

Based on the project Fast-Planner.