matthias-mayr/skiros2

A skill-based platform for ROS

SkiROS2: Skill-based robot control platform for ROS

Overview

SkiROS2 is a platform to create complex robot behaviors by composing skills - modular software blocks - into behavior trees.

Robots coordinated with SkiROS can be used in partially structured environments, where the robot has a good initial understanding of the environment, but it is also expected to find discrepancies, fail using initial plans and react accordingly.

SkiROS offers the following features:

• A framework to organize the robot behaviors within modular skill libraries
• Scalable skill model with pre-, hold- and post-conditions
• A reactive execution engine based on Behavior trees
• A world model as a semantic database to manage environmental knowledge
• Reasoning capabilities and automatic inference of skill parameters
• An integration point for PDDL task planning
• Automatic generation of planning domains based on the skills and entities in the world model
• ROS, RViz and tf integration
• Python APIs for skill handling, the world model and task planning

Getting Started

• A full introduction and the tutorials are located in the wiki
• Executable skill examples are in the skiros2_examples repository
• The SkiROS2 paper provides an overview and background information
• The skiros2_template_lib provides a skeleton for new skill library
• Installation instructions are below

Installation

Compatibility

SkiROS is compatible with Ubuntu 18.04/ROS Melodic and Ubuntu 20.04/ROS Noetic, Python 2 and 3.

Installation Instructions

To use SkiROS you must have ROS installed on your machine. You also need pip to install python dependencies.

Clone this repository into your catkin workspace src directory:

mkdir -p catkin_ws/src && cd catkin_ws/src
git clone https://github.com/RVMI/skiros2
# Clone the base skill set into the skiros2 directory in your catkin workspace.
git clone https://github.com/RVMI/skiros2_std_lib
# Optionally, you can clone the skiros2_examples repositories here as well:
git clone https://github.com/RVMI/skiros2_examples

Install dependencies defined in each package.xml using rosdep and the python dependencies:

rosdep install --from-paths . --ignore-src --rosdistro=$ROS_DISTRO -y
# Install Python dependencies
pip install -r requirements.txt --user

Build the workspace with catkin

cd ~/catkin_ws
catkin_make # or 'catkin build'
source ./devel/setup.bash

Launch SkiROS2

roslaunch skiros2 skiros.launch
# Or try one of the examples like
roslaunch skiros2_examples turtlesim_example.launch

Task Planning

Optionally, if you want to use the task planning skill provided in the standard library, install the fast downward planner with the following script:

cd skiros2/scripts
./install_fd_task_planner.sh

When asked for the install folder, you can insert a location of your preference or just leave the default.

Creating own skills

The skiros2_template_lib provides the necessary structure to create a new library from scratch. It also has an example launch file to start the system.

Build the Doxygen documentation

To generate the doxygen documentation, first install Doxygen and doxypypy for better compatibility with python doc-strings:

sudo apt-get install -y doxygen
pip install doxypypy --user

Then, in the top level directory, generate the documentation and view it with a browser, e.g. Firefox:

doxygen Doxyfile
firefox skiros2_doc/html/index.html

Citation

A preprint of the SkiROS2 paper is available here. If you are using it in your work, we would be pleased if you would cite it:

@misc{mayr2023skiros2,
      title={SkiROS2: A skill-based Robot Control Platform for ROS}, 
      author={Matthias Mayr and Francesco Rovida and Volker Krueger},
      year={2023},
      eprint={2306.17030},
      archivePrefix={arXiv},
      primaryClass={cs.RO}
}

Acknowledgements

This platform has been developed at the RVMI lab.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 723658, Scalable4.0.

This work was partially supported by the Wallenberg AI, Autonomous Systems and Software Program (WASP) funded by Knut and Alice Wallenberg Foundation