The League of Robot Runners
The inaugural League of Robot Runners is a competition series that
tackles one of the most challenging problems in automated logistics:
coordinating a team of moving robots.
Simplified versions of such problems appear in the research literature
under the name Multi-Agent Path Finding (MAPF), Multi-Robot Path
Planning (MRPP), Multi-Agent Coordination, and many others. Sponsored by
Amazon Robotics, the League is designed to bridge the gap between
academia and industry, by identifying the core combinatorial challenges
that make robot coordination problems difficult to solve in practice.
In the competition, a team of (simulated) robots work together to
complete errands. They fulfil orders, sort mail and navigate tricky grid
environments. Participants must compute efficient and collision-free
paths for the team. By following these paths the robots can get to where
they're going as quickly as possible, so that they can complete as many
errands as possible, all before time runs out.
The competition is open to teams and individuals from any geographic
region and from any community. All approaches (and their combinations)
are welcome, including but not limited to:
- algorithmics and AI planning,
- combinatorial search and optimisation,
- machine learning and reinforcement learning,
- mathematical programming and operations research, and
- any other class of reasoning techniques.
Awards and Prizes:
Participants will test their skills against the best in the field,
competing for fame and financial prizes across three categories:
Category 1; Best overall: $5,000 USD
The winner of this category has the highest overall score across all
problem instances in the competition.
Category 2; Line honours: $2,500 USD
The winner of this category has the largest number of best solutions
across all problem instances in the competition.
Category 3; Fast mover: $2,500 USD
Participants are eligible for this category if their solver always
returns a set of valid actions (one for each agent) before the end of
each planning episode (one second each). The winner of this category has
the highest overall score among all eligible participants.
Notable performances in each category (2nd, 3rd) will be highlighted and
recognised with an award certificate.
A limited number of training awards are also available on a basis of
merit. These awards come in the form of AWS credits worth $1,000 USD and
are intended to help participants who need large computational resources
for pre-computations (for example, model training).
Submission Process:
Participants register on our website with their GitHub credentials.
Competing solvers are then uploaded to a private repository. The
submitted code is automatically cloned, compiled, and executed on our
evaluation server, and participants can track their progress via our
live leaderboard. Submissions are welcome anytime before the deadline!
We provide starter code, example problems, solution validators,
visualisers, and extensive documentation. After the competition,
participants are required to open source their submissions so they can
be studied and adopted by others in the research community.
Important dates:
- November 30th, 2023: Competition ends
- December 20th, 2023: Results are announced
Competition chairs:
- Daniel Harabor (Monash University)
- Sven Koenig (University of Southern California)
- Cathy Wu (Massachusetts Institute of Technology)
- Jingjin Yu (Rutgers University)
For more information,please visit our website:
www.leagueofrobotrunners.org
For community discussion,join our official Discord
https://discord.gg/CEYT4g4raR
We hope to see you there!
**********************************************************
*
* Contributions to be spread via DMANET are submitted to
*
* DMANET@zpr.uni-koeln.de
*
* Replies to a message carried on DMANET should NOT be
* addressed to DMANET but to the original sender. The
* original sender, however, is invited to prepare an
* update of the replies received and to communicate it
* via DMANET.
*
* DISCRETE MATHEMATICS AND ALGORITHMS NETWORK (DMANET)
* http://www.zaik.uni-koeln.de/AFS/publications/dmanet/
*
**********************************************************
Posts
