This page is no longer kept up to date. EISBot information can now be found here.
EISBot is the Expressive Intelligence Studio's entry for the upcoming StarCraft AI Competition to be held at the 2010 Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE). The bot uses a reactive planning language and will include several forms of machine learning. It interacts with StarCraft using the BWAPI project.
Developing agents capable of defeating competitive human players in real-time strategy games remains an open research problem. The EISBot project aims to achieve this goal and in undertaking this process, address several interesting AI research challenges. Notably, building systems capable of intelligently interacting with human participants in complex, real-time, partially observable domains in which actions must be concurrently executed at multiple scales. The goal of achieving human-level performance in this domain will be accomplished by integrating several forms of domain knowledge including hand-authored behavior, expert demonstrations, and player created repositories of strategies and tactics. To employ these disparate sources of knowledge, EISBot will utilize an ensemble of symbolic and statistical AI techniques.
A video showing off some of the debug features in the EISBot:
- Ben Weber and Michael Mateas. A Data Mining Approach to Strategy Prediction, IEEE Symposium on Computational Intelligence in Games (CIG 2009).
- Ben Weber, Michael Mateas and Arnav Jhala. Applying Goal-Driven Autonomy to StarCraft, Artificial Intelligence and Interactive Digital Entertainment (AIIDE 2010).
- Ben Weber and Santiago Ontañón. Using Automated Replay Annotation for Case-Based Planning in Games. ICCBR Workshop on CBR for Computer Games (ICCBR-Games 2010).
- Ben Weber, Peter Mawhorter, Michael Mateas and Arnav Jhala. Reactive Planning Idioms for Multi-Scale Game AI. IEEE Conference on Computational Intelligence and Games (CIG 2010),
- Ben Weber, Michael Mateas and Arnav Jhala. Case-Based Goal Formulation. AAAI 2010 Workshop on Goal-Driven Autonomy (GDA 2010).
Country map of ICCup users versus the EISBot:
You can download and try out the initial version of the EISBot.
A new version is available here: EISBOT 2.6.1
- To install: copy ExampleAIModule to bwapi-data/AI
- To run: launch "run client.bat"
- Select Terran as the bot's race
- Download here