RoboCup Logistics League

Introduction

In the RoboCup Logistics League (RCLL) teams of three robots need to organize and maintain the supply chain in a simplified factory to produce several product variations. It covers many aspects of mobile robotics research and the field of Industry 4.0.

The Carologistics RoboCup Team participates in the RCLL as a joint effort of the Knowledge-Based Systems Group (RWTH Aachen University), the Cybernetics Lab IMA/ZLW & IfU (RWTH Aachen University), and the MASCOR Institute (FH Aachen UoAS) since 2012. The team has won the competition 2014—2017.

Carologistics YouTube Channel.

The RCLL is part of RoboCup Industrial, an effort to closely coordinate with the second industrial application league RoboCup@Work. The following video by the RoboCup Federation gives an in-depth overview over RoboCup Industrial.

The Challenge

The teams have to maintain and optimize the material flow in a virtual factory. The products are stacked colored bases and rings, according to orders which become known only at run-time. The robots must then generate a plan and execute it, considering uncertainty and yet optimizing it, e.g., by minimizing travel distances. Each team has a set of machines which serve specific purposes, such as providing product base elements or mounting colored rings and caps. Teams score for successfully delivered products. Two teams compete at the same time. While each team has a set of machines for their exclusive use, requiring collision avoidance and re-routing if a passage is busy with a robot from another team increases uncertainty. Robots can communicate by a wireless network, however, especially during RoboCup events, availability and reliability of the network can vary, making robust coordination a necessity.

Research

The RCLL provides a testbed and application domain in particular for research on multi-robot task planning and coordination with robust execution. Teams must fulfill basic requirements for robotics systems, such as self-localization and navigation, but requirements in terms of perception and manipulation are kept intentionally simple, in order to focus on planning and reasoning. In addition, the RCLL provides a relevant and industry-inspired testbed. This fits our general line of research.

We focus on knowledge representation and reasoning, here in the form of multi-robot planning, world model synchronization, and robust coordination and execution.

Selected Publications

Our Involvement

We have participated in the RCLL since 2012. We had major influence on the advancement and development of the league with members filling roles on the Executive, Technical, and Organization Commitees.

Team

The team consists mostly of PhD and master students of Computer Science and Mechanical Engineering. Team leaders were (in chronological and alphabetical order) Daniel Ewert (2012-2013), Alexander Ferrein (2012), Tim Niemueller (2012-2017), Sebastian Reuter (2012-2015), Tobias Neumann (member since 2014, team leader 2016-2017), Christoph Henke (since 2017), Till Hofmann (since 2017), Victor Matare (member since 2012, team leader 2017), Nicolas Limpert (member since 2014, team leader since 2017), Sebastian Schönitz (since 2017).

The team is supervised by Prof. Dr. A. Ferrein, Prof. Dr. S. Jeschke und Prof. G. Lakemeyer, PhD.


In addition to the team leaders and supervisors, team members were David Bosen (since 2018), Andre Burghof (2012-2013), Sebastian Eltester (since 2018), Mostafa Gomaa (since 2016), Daniel Habering (since 2017), Daniel Künster (2016-2017), Randolph Maaßen (2014-2015), Matthias Löbach (2014-2016), Bahram Maleki-Fard (2013), David Masternak (2015), Florian Nolden (2013-2014), Johannes Rothe (2013-2016), David Schmidt (2016-2017), Richard Schulz (2013), Morian Sonnet (since 2018), Carsten Stoffels (since 2017), Tarik Viehmann (since 2018), Alexander von Wirth (2013), Frederik Zwilling (2013-2016).

Fawkes Robot Software Framework

We have made our full software stack based on Fawkes publicly available since 2014 with a release every year (2014, 2015, 2016). It has been adapted by other teams. We were awarded the 3rd place in the 1st International Harting Open Source Award in 2016.

Planning and Execution Competition in Simulation

Based on our software stack and Gazebo-based simulation, we have partnered with the Model-based Embedded & Robotic Systems Group from MIT and Cognitive Robotics Lab at Technion Israel to run the Planning and Execution Competition for Logistics Robots in Simulation in 2017 and 2018 at the ICAPS conference.

Outreach

We have hosted yearly Hackathons for students from RWTH Aachen University and FH Aachen UoAS. Most team members have joined after participating in such events.

We have organized several winter and summer schools with the RCLL as a means to teach students the relevant methods.

Awards

Awards won by the Carologistics team related to the RoboCup Logistics League.