The 16th edition of the Multi-Agent Programming Contest

Publication

The Scenario: Agents Assemble III

Agents with limited local vision have to organize to assemble and deliver complex structures made of blocks, in a grid world.

The scenario is a revision of the previous editions, 2019 and 2020/21 .

Also see the official Call for Participation.

Contest

Results

Replays

Wednesday (21st September 2022)

Thursday (22nd September 2022)

Warmup matches

Each team played warmup matches against master student Paula Böhm from TU Clausthal, outside of the competition.

Match schedule

Wednesday (21st September 2022)

Thursday (22nd September 2022)

Configuration

The following parameters were used in the contest matches:

• Sim 1
• Sim 2
• Sim 3

Timeline

Replanned schedule

Due to popular demand, the dates have been moved into late(r) 2022.

Participants

In order of registration:

Registration

To register, submit the registration form.

Qualification

In order to participate in the contest, your agents have to demonstrate a stable connection and some basic scenario-related skills. Details will be made available closer to the contest.

Connection Tests

After the registration deadline, you can contact us to schedule a connection test, where you can try out your connection to our severs (and possibly identify some issues before the qualification).

Anonymous Test Matches

You can tell us if you’d like to play test matches against other teams (before the contest!). If we find a match, both teams’ identities will be kept secret from each other. The full monitor will of course also be disabled.

The platform

The MASSim server is available on GitHub.

The package contains some dummy agents for different platforms. You can use them as a starting point or implement the communication protocol yourself.

Prize

The winner of the contest will be awarded a voucher for 500 EUR worth in books, thankfully provided by Springer Verlag. Requirements are the submission of a team description paper and the source code of the agents.

Communication

Mailing List

Participants and all interested colleagues are invited to subscribe to our mailing list:

agentcontest@googlegroups.com

All the important details and announcements will be shared on this list.

The mailing list archive is publically available.

Slack

You can also join our Slack community for discussions, questions, bug reports, etc.

Publications

After the tournament we invite every participant to submit a paper about their team. Once the quality of the papers has been assured, they will be regularly published (as the previous editions) in Springer’s LNCS Challenges subline.

Authors should consult Springer’s authors’ guidelines and use their proceedings templates for the preparation of their papers. Springer encourages authors to include their ORCIDs in their papers. In addition, the corresponding author of each paper, acting on behalf of all of the authors of that paper, must complete and sign a Consent-to-Publish form. The corresponding author signing the copyright form should match the corresponding author marked on the paper. Once the files have been sent to Springer, changes relating to the authorship of the papers cannot be made.

• Strict submission deadline: 18 December 2022
• Submission system: EquinOCS (click the small Enter Service button)

Please answer and append the questionnaire to your team description paper. Also, please send us your questionnaire responses (only!) until 14 October 2022.

General Advice

We have collected some tips and tricks for participating in the contest. If you have participated before, please send us your additions.

• Make sure your agents can handle transitions between simulations.
• During the contest, only the status monitor will be active. Maybe enable your agents to tell you what’s going on.
• Often, we see agents reconnecting to the server during a simulation. Maybe try to ensure that:
  • your agents do not lose valuable information if they need to be restarted; and
  • they can handle different initial states (e.g., agent is carrying a block).
• Test your agents against other agents (e.g. a second instance of your team).
• Test your agents with different scenario parameters.
• Use the skip() action for each agent who needs to wait a step in order to speed up the simulation.

For the current scenario:

• If an agent is somehow attached to another, it moves together with the other agent. These agents can coordinate their movements to move faster! (One move-action per agent and step.)
• Your agents can start or end up surrounded by obstacles. Make sure they know how to get out of such situations.