• Newsletters
  • Submit News
  • SCLegacy RSS

Over the years, StarCraft has been a primarily Multiplayer-driven game. The Sci-Fi RTS and its community have both been kept alive by the spirit of competition between professional players all over the world and the love of their own fans. The game's single-player experience has since captivated others over the years, because of its compelling Sci-Fi story and lore.

Beyond this, the game has also provided a private arena for its most casual fans ever since its initial release. This arena is known by many player as the art of good old fashioned comp stomping. Most players would challenge themselves to see how many computer opponents they could take on at once, simply because the AI was as crude and lackluster as you would expect from a Legacy game such as StarCraft. A pair of PhD students hope to remedy this by developing an intelligent AI opponent that not only plays on a much higher level, but also evolves and learns build orders by evaluating data from others' replays.

This short clip showcases the new AI (Blue Terran) fighting against a human player by laying spider mines on top of a bridge out the human player's sight. The AI's developers not only want to see if their AI, EISBot, can make the default StarCraft AI obsolete, but they also want to prove that it can compete against the best StarCraft pro-gamers in the world.

Real-time strategy games provide an excellent environment for A.I. research and creating bots that are capable of defeating skilled players in this domain is still an open problem. EISBot is the Expressive Intelligence Studio's StarCraft bot and is part of our dissertation research. It is coded in a reactive planning language and is composed of managers that handle different aspects of gameplay. EISBot selects build orders from a set of replays using case-based reasoning. Our goal is to build a bot that learns how to play StarCraft competitively based on analysis of expert StarCraft replays.

Even though the sample clip above shows EISBot playing TvT, the bot plays Protoss as its race of choice. It currently has a healthy 20% win ratio against human players this early in development.

Click here to read more.


Ventrilo Servers
Contact Us About Us

SCLegacy is hosted by DarkStar Communications, home to high quality Ventrilo Servers.