Implementation of Jump Point Search with two different back-end graph search libraries: YagSBPL and NX lib. Original jump point seach algorithm is proposed in "D. Harabor and A. Grastien. Online Graph Pruning for Pathfinding on Grid Maps. In National Conference on Artificial Intelligence (AAAI), 2011". The 3D version is proposed in "S. Liu, M. Watterson, K. Mohta, K. Sun, S. Bhattacharya, C.J. Taylor and V. Kumar. Planning Dynamically Feasible Trajectories for Quadrotors using Safe Flight Corridors in 3-D Complex Environments. ICRA 2017".
Required:
- Boost
- Eigen
- yaml-cpp (Reading data files)
- VTK (For visualizing output)
A) Simple cmake
$ mkdir build && cd build && cmake .. && makeB) Using CATKIN with ROS
$ mv jps3d ~/catkin_ws/src
$ cd ~/catkin_ws & catkin_make -DCMAKE_BUILD_TYPE=ReleaseNote that in other repository, add following command in CMakeLists.txt in order to correctly find jps3d:
include_directories(${JPS3D_INCLUDE_DIRS})The simple API are provided in the base planner class, here are some important functions to set up a planning thread:
std::unique_ptr<PlannerBase> planner(new XXXUtil(false)); // Declare a XXX planner
planner->setMapUtil(MAP_UTIL_PTR); // Set collision checking function
bool valid = planner->plan(start, goal); // Plan from start to goalIn this library, we consider 3D voxel grid but user can write their own 2D map util plugin using the MapBaseUtil class. Two planners are provided as follows:
GraphSearch2DUtilGraphSearch3DUtil
The results from A* and JPS are plotted in corridor.jpg.
Green path is from A*, black path is from JPS.
$ ./build/test_planner_2d ../data/corridor.yaml
start: 2.5 -2 0
goal: 35 2.5 0
origin: 0 -5 0
dim: 799 199 1
resolution: 0.05
JPS Planner takes: 4.000000 ms
JPS Path Distance: 35.109545
AStar Planner takes: 77.000000 ms
AStar Path Distance: 35.109545For more details, please refer to https://sikang.github.io/jps3d/index.html