From the Starcraft AI Competition Webpage: "Computer programs will play Starcraft Brood War games against each other using the BWAPI, a software library that makes it possible to connect programs to the Starcraft: Brood War game engine. The purpose of this competition is to foster and evaluate progress of AI research applied to real-time strategy (RTS) games. RTS games pose a much greater challenge for AI research than chess because of hidden information, vast state and action spaces, and the requirement to act quickly."
For my Game AI class, we used the Starcraft Brood War API to implemented a behavior tree and blackboard for unit micromanagement. Our goal for the class was to defeat the default Starcraft AI in a 12 v. 12 Dragoon skirmish.
We had two primary concerns for our Dragoon Squad:
- Try to form the best possible concave before engaging the enemy.
- Concentrate fire on enemy Dragoons that can be attacked by friendly units.
- Retreat any units that are low health.
In the video below, the units first re-position themselves before combat. Once combat is entered, units will attempt to retreat if a calculated retreat is possible.
This sparked my interest to further pursue game AI by joining my school's Starcraft AI Team, Cruzbot. The Cruzbot team used an early version of the UAlbertaBot AI program to explore alternative heuristics for the bot's alpha-beta search and updated build order strategies.