(please note that the following descriptions have been provided by the bot-makers)

Florian Richoux, University of Tokyo:

The main idea of the Artificial Intelligence Using Randomness (AIUR) is to be unpredictable by making some stochastic choices. The AI starts a game with a “mood” (randomly picked up among 5 moods), dictating some behaviors (aggressive, defensive, macro-game, …). In addition, some other choices (productions, timing attacks, early aggressions, …) are also taken under random conditions.

Johan Hagelbäck, Blekinge Institute of Technology:

BTHAI is based on a general, easily expandable multi-agent architecture which allows it to play reasonably well as Terrans, Protoss or Zerg. Buildorders, squad setups and research orders are defined in scripts which makes it very easy to completely change strategy without a recompile of the bot.

Yujing Hu, University of Nanjing:

EvoBot uses various AI method(e.g. evolutionary algorithm for obtaining rational unit combinations, influence map techniques for deciding the strategic locations, opponent modeling for exploiting the weakness of enemies). Now we have only completed the work of influnce map and some basic functions (such as training units, building constructions).

Alberto Uriarte, IIIA-Spanish National Research Council:

Nova is a multi agent system that simulates a human player. The agents use different AI techniques such us: a blackboard architecture, a working memory, FSM, threat maps, potential fields, group behaviours, reactive planning, …

Andrew Smith, freelancer:

Skynets main features include:

• A fast custom terrain analyser.
• An advanced building placer that creates tight but (mostly) non blocking bases.
• A task based macro system that continually plans and fully understands all requirements.

Gabriel Synnaeve, E-Motion team at INRIA Rhône-Alpes (LIG) / University of Grenoble:

BroodwarBotQ uses probabilistic techniques both for micro management and strategy. A Bayesian model learned from high skill player is used to determine the opponent's strategy and a Bayesian sensory motor fusion model is used for micro-management.

Douglas Patti and Matthew Taylor, Lafayette College:

LSAI is a newly developed Zerg bot on a tight time constraint with limited resources. It utilizes a heavily modified BWSAL to divide management of the units to different modules that communicate via a centralized information module. It works using a simple reactive strategy to try and survive early game attacks and macro up to a larger attack force and maintain map control.

Joshua Dong and Randall Blake, Westwood High School:

Due to time constraints and low team-member count in addition to our relative inexperience with C++ we were not able to program very quickly, but we had some basic Starcraft-playing experience, and thus knew a few tricks to exploit the execution speed of an AI script. This is why we chose Protoss, as we knew there were ways to exploit the strength of probes in early-game PvT. Protoss was also favorable due to the sheer strength of zealots in early-game in PvZ. We had many good ideas but could only implement a few, and to varying degrees of quality. Thus, we were programming a bot that won tactically, based on stronger mechanics and exploitation of the computer advantage. We knew we did not have time to implement a good strategical response, thus chose to use an all-in strategy that instead forced the opponent to react correctly or risk a large disadvantage. We felt that a 5-gate Zealots build would suffice fro this, as it counters any tech with the threat of elimination outright, and is quite strong against any other race's tier-one units. Normally in the early game situation one base can only support 4 gateways, but due to our decision not to use tech and only make zealots, coupled with our improved resource gathering strategy, we were able to quickly support a 5 gate all-in zealot rush. The bot should be stronger on small maps and maps for 2 players, as this forces the opponent to react quickly, narrowing their options.

Using the strong mechanics and multi-tasking ability of the computer, we are able to gather resources at a faster rate whilst using fewer workers than any human – our bot only needs 21 probes for full base saturation, or 2 workers per mineral patch and 3 for each geyser. Our bot does not use gas though currently, so it only requires 18 workers on minerals to acheive 100% saturation. Each mineral group is eventually assigned a pair of workers that will only mine from that mineral until an expansion has been setup and the mineral patch has been mined out.

Some current points to be resolved or developed in the future:

• Scouting and reacting
• Expanding
• Choosing a tech route
• Improving code efficiency and effectiveness

Ho-Chul Cho and Kyung-Joong Kim, Sejong University, Seoul:

Starcraft is one of the most popular games in Korea. There are pro-gamers and even proleagues for the game. I have also played starcraft for a long time and gained lots of information about it.

Based on my abundant infomation about the game, i have found not only the strategies for winning but also cunning tactics that hardly require human intelligence.

Therefore, the bot is determined to achieve the goal and the programming code is simple. It generally uses a rule base from artificial intelligence and is an expanded version of Aiurbot (old version) based on BWSAL.

David Churchill, University of Alberta, Canada:

A Protoss bot which uses early and constant pressure to contain or outright kill its enemy. Build orders are planned and implemented in real-time via depth-first branch & bound heuristic search.

back to the competition main page