ROS implementation of a global 2D path planner based on state lattices.
The package has been developed and tested against the ROS Kinetic distro.
The matio library is required for loading lattice parameters from .mat files. Until this PR is merged, it must be installed manually:
sudo apt install libmatio-devFurthremore, the package depends on agv_control_msgs. Download agv_control_msgs into the same catkin workspace before building this package.
The pathPlanner package provides path planning functionality, based on the state lattice approach, as described in Draganjac et. al. 2016, Decentralized Control of Multi-AGV Systems in Autonomous Warehousing Applications.
Subscribed topics: /tf (robot localization information), /goal (goal point)
Published topics: /plan (the path planned by the robot)
WARNING: Due to a limitation in the current planner implementation, it requires the map origin to be at [0,0,0], otherwise the planner will fail to find feasible paths.
To test the node use the provided test.launch file. You can set the pose of a dummy robot in RViz with the 2D Pose Estimate tool and you can provide navigation goals with the 2D Nav Goal tool. Before setting the first planning goal, make sure to set a pose estimate outside of the map edges, otherwise planning will fail (map edges are treated as obstacles by the planner). For distances greater than 10 meters, the planning can take quite a long time.
If you find this package useful, please consider citing our work:
@ARTICLE{7571170,
author={I. Draganjac and D. Miklić and Z. Kovačić and G. Vasiljević and S. Bogdan},
journal={IEEE Transactions on Automation Science and Engineering},
title={Decentralized Control of Multi-AGV Systems in Autonomous Warehousing Applications},
year={2016},
volume={13},
number={4},
pages={1433-1447},
keywords={},
doi={10.1109/TASE.2016.2603781},
ISSN={1545-5955},
month={Oct},}