Skip to content

Latest commit

 

History

History
145 lines (102 loc) · 5.18 KB

README.md

File metadata and controls

145 lines (102 loc) · 5.18 KB
Raw

Multi-Robot Coverage Planning

Software stack for Cooperative Multi-Robot Coverage for Cleaning Robots.

Developed for : AIITRA Robotics Challenge 2021 - Team Curious Comrades (Winner!)

ros-ci lint

Developed and Tested on ROS Noetic + Ubuntu 20.04 + Gazebo 11

Features

  • Solution for autonomous Multi-Robot SLAM using Frontier based exploration
  • Optimizer for finding optimal number of agents required to cover an occupancy grid
  • Coverage path planner based on Boustrophedon Cellular Decomposition
  • PID path tracking

Build

Clone the project

git clone https://github.com/nocoinman/Multi-Robot-Coverage-Planning.git
cd MRCP

Resolve dependencies using rosdep

rosdep install -y -i --from-paths ./src

Build workspace

catkin_make --cmake-args -DCMAKE_BUILD_TYPE=Release

Run the tests to make sure everything is setup correctly (optional)

catkin_make run_tests && catkin_test_results build/test_results

Usage

Shell 1

source devel/setup.bash
roslaunch simulation multi_robot.launch map:=map1

Available maps: map1, map2, map3, map4

Shell 2

source devel/setup.bash
roslaunch full_coverage_path_planner cover_map.launch map:=map1

Unpause physics to start simulation

Gazebo by default starts in headless mode, with physics paused. You'll have to unpause physics to start the simulation. You can do this in 2 ways.

For the shell geeks out there, you can unpause physics from the shell using rosservice

rosservice call /gazebo/unpause_physics

For the GUI fellas, we have conveniently added a custom panel to RViz to pause and unpause gazebo physics. Thank us later!

Workflow

Intended use of the software is as shown.

Coverage Path Planning

Coverage path is computed using Boustrophedon Cellular Decomposition. The plan is then divided into equal sub-plans and assigned to each agent.

Here is a demo of the path planning algorithm in action, ran on the occupancy grid generated using Multi-Robot SLAM on a complex office setting.

Path Tracking

Agents follow the designated path asynchronously. This is accomplished using simple PID control based path tracking.

Acknowledgements

The following open-source projects were really insightful for the implementation of this project. These were improvised and adapted for our use case. Please check them out!

topics_rviz_plugin

full_coverage_path_planner

tracking_pid

The following research papers and articles were referred during ideation and development.

@article{DBLP:journals/corr/abs-1806-03581,
  author    = {Anirudh Topiwala and
               Pranav Inani and
               Abhishek Kathpal},
  title     = {Frontier Based Exploration for Autonomous Robot},
  journal   = {CoRR},
  volume    = {abs/1806.03581},
  year      = {2018},
  url       = {http://arxiv.org/abs/1806.03581},
  eprinttype = {arXiv},
  eprint    = {1806.03581},
  timestamp = {Mon, 13 Aug 2018 16:46:39 +0200},
  biburl    = {https://dblp.org/rec/journals/corr/abs-1806-03581.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

@InProceedings{10.1007/978-1-4471-1273-0_32,
    author="Choset, Howie
    and Pignon, Philippe",
    editor="Zelinsky, Alexander",
    title="Coverage Path Planning: The Boustrophedon Cellular Decomposition",
    booktitle="Field and Service Robotics",
    year="1998",
    publisher="Springer London",
    address="London",
    pages="203--209",
    abstract="Coverage path planning is the determination of a path that a robot must take in order to pass over each point in an environment. Applications include vacuuming, floor scrubbing, and inspection. We developed the boustrophedon cellular decomposition, which is an exact cellular decomposition approach, for the purposes of coverage. Each cell in the boustrophedon is covered with simple back and forth motions. Once each cell is covered, then the entire environment is covered. Therefore, coverage is reduced to finding an exhaustive path through a graph which represents the adjacency relationships of the cells in the boustrophedon decomposition. This approach is provably complete and Experiments on a mobile robot validate this approach.",
    isbn="978-1-4471-1273-0"
}