Graph-based Exploration Planner 2.0

Tutorial:

Please refer to the wiki page for detailed instructions to install and run the demo simulations as well as documentation of the planner interface and parameters. More results and video explaination of our method can be found on our website: Link

Installation

These instructions assume that ROS desktop-full of the appropriate ROS distro is installed.

Install necessary libraries:

For Ubuntu 18.04 and ROS Melodic:

sudo apt install python-catkin-tools \
libgoogle-glog-dev \
ros-melodic-joy \
ros-melodic-twist-mux \
ros-melodic-interactive-marker-twist-server

For Ubuntu 20.04 and ROS Noetic:

sudo apt install python3-catkin-tools \
libgoogle-glog-dev \
ros-noetic-joy \
ros-noetic-twist-mux \
ros-noetic-interactive-marker-twist-server

Create the workspace:

mkdir -p gbplanner2_ws/src/exploration
cd gbplanner2_ws/src/exploration

Clone the planner

git clone git@github.com:ntnu-arl/gbplanner_ros.git -b gbplanner2

Clone and update the required packages:

cd <path/to/gbplanner2_ws>
wstool init
wstool merge ./src/exploration/gbplanner_ros/packages_ssh.rosinstall
wstool update

Build:

catkin config -DCMAKE_BUILD_TYPE=Release
catkin build

Running Planner Demo

Aerial Robot Demo

Download the gazebo model from here and extract in the ~/.gazebo/models folder.

roslaunch gbplanner rmf_sim.launch

It will take few moments to load the world. A message saying the spawn_rmf_obelix process has died may pop up, but as long as the pointcloud map is visible in rviz and the UI controls work this message can be safely ignored.

Ground Robot Demo

the following command:

roslaunch gbplanner smb_sim.launch

In Ubuntu 18.04 with ROS Melodic, the gazebo node might crash when running the ground robot simulation. In this case set the gpu parameter to false here.

Results

Robot's of Team Cerberus running GBPlanner and GBPlanner2

Autonomous exploration mission in the Prize Round of the DARPA Subterranean Challenge Final Event using four ANYmal C legged robots (Chimera, Cerberus, Camel, Caiman), all running GBPlanner2 independantly.

References

Explanation Video

If you use this work in your research, please cite the following publication (link)

@article{dang2020graph,
  title={Graph-based subterranean exploration path planning using aerial and legged robots},
  author={Dang, Tung and Tranzatto, Marco and Khattak, Shehryar and Mascarich, Frank and Alexis, Kostas and Hutter, Marco},
  journal={Journal of Field Robotics},
  volume = {37},
  number = {8},
  pages = {1363-1388},  
  year={2020},
  note={Wiley Online Library}
}

You can contact us for any question:

TurtleZhong / gbplanner_ros