Realtime Strategy (RTS) games are one of the major computer game genres and one of the few for which AI-based players (bots) have little chance to win against expert human players — if they are not allowed to cheat. We presume that competitions as this one will help to improve AI development in the RTS genre. StarCraft (by Blizzard) is one of the most popular RTS games of all time, is known to be extremely well balanced, and thus is an ideal candidate to test different AI approaches.

This competition is part of the Computational Intelligence in Games conference 2011. This year, we have two tracks, please let us know where you intend to start (of course you can do both).

For both tracks we will hold round-robin tournaments with 10 games per match and fixed time limit of 1h (to prevent endless games after resources are used up) after which the built-in StarCraft score decides the winner. The bot with the highest overall number of single-game victories is the winner of the track. See below for more info.

A PDF competition poster is here.

Tournament is finished, see results below.

If you have any questions or comments, please let us know via mail starcraft@ls11.cs.tu-dortmund.de.

(see an extended description of the bots here:)

Tournament is finished, see below for results.

Unfortunately we have received only one submission for Track B. We therefore decided to cancel the track. Johan, we are very sorry.

As all games have been played manually, we decided to set up 2 brackets with 5 bots each on a random basis. From each bracket, the 2 best bots make it into the final round (counting the number of wins, and evaluating the direct encounters only in case of equal wins). To compensate for slight hardware differences, each pair has played two games (one on each of the two employed machines), and this has been repeated for all 5 maps used in the first round. We are aware the fact that the split into two groups is not entirely fair (as seen for the example of AIUR who won more games than the 2nd bot in the other bracket) but had to reduce the number of overall played games.

Qualified for the final round: UAlbertaBot and Skynet.

As BroodwarBotQ and BTHAI have the same number of wins, their direct encounter is evaluated which accounted 6 : 4 for the BroodwarBotQ.

Qualified for the final round: Xelnaga and BroodwarBotQ. Note that all qualified bots play Protoss.
Maps used for the first round: (2)MatchPoint 1.3, (4)Fighting Spirit 1.3, iCCupdestination 1.1, iCCup gaia, iCCup great barrier reef

The final round was played in a similar mode as each of the first round brackets, using another set of 5 maps (see below), resulting in 30 games for each bot.

This makes Skynet (Andrew Smith) the clear winner, followed by UAlbertaBot (David Churchill) on 2nd and Xelnaga (Ho-Chul Cho and Kyung-Joong Kim) on 3rd place. Congratulations!

Maps used in the final round: iCCup lost temple 2.4, iCCup rush hour 3.1, iCCup swordinthemoon2.1, iCCup yellow 1.1, La_Mancha1.1

These presentations have been given by Mike and Tobias at CIG 2011 in Seoul on September 2, the first round at 9:20 AM, the final round at 14:25 PM. Powerpoint and PDF have the same contents, but the sound is only included in the Powerpoint slides.

First round Powerpoint / First round PDF

Final round Powerpoint / Final round PDF

We will most likely hold a similar tournament next year again (however, there may be slight rule changes), so you are invited to prepare your bots. In this case, you may follow these instructions:

The competition will use StarCraft Brood War 1.16.1. You must possess a legal copy of the game.

Bots for StarCraft shall use the Broodwar API, which provides hooks into StarCraft and enables the development of custom AI for StarCraft. Interfaces in many languages are available to query the current state of the game and issue orders to units. Note that due to an important bug release we decided to use BWAPI 3.6.1 as also employed for the AIIDE tournament.

• Instructions for setting up the environment are available here.
  
• An introduction to the Broodwar API is available here.
  
• Instructions for building a bot that communicates with a remote process are available here.
  
• A new JNI (Java) bridge to the BWapi by Ben Weber is available here.
  

Specific (tentative) ruleset for track A (this is largely identical to the AIIDE ruleset).
Specific ruleset for this track B of the competition (nearly unchanged from last year).
Baseline (tentative) ruleset (the same as for the AIIDE competition, it basically defines what misbehavior of a bot is that would lead to disqualification).

We will largely use a pool of maps widely used for internet gaming, as e.g. the maps of the International Cyber Cup, thus we recommend using some of these also for training your bots.

The 2nd Annual AIIDE Starcraft AI Competition

Tobias Mahlmann, IT University Copenhagen
Mike Preuss, Technische Universität Dortmund

For any comments or inquiries, please write to starcraft@ls11.cs.tu-dortmund.de

StarCraft®: Brood War®
Used with permission granted to CIG. Blizzard has consented to run the competition via the Broodwar API interface, which uses a back door of the game and is not in any form related to Blizzard. Thanks!

©1998 Blizzard Entertainment, Inc. All rights reserved. StarCraft, Brood War and Blizzard Entertainment are trademarks or registered trademarks of Blizzard Entertainment, Inc. in the U.S. and/or other countries.