TD Learning for Build Order Implemented and Trained

This release contains our new Temporal Difference learning algorithm that determines the bots build order. This algorithm using machine learning to determine the best next move given a current state of the game.

Genetic Algorithm Fully Trained

This release contains our previously implemented Genetic Algorithm for building placement and is the most trained version that our team will produce.

Genetic Algorithm Partially Trained

This release includes the genetic algorithm implemented in the last release with the addition that it has been trained on maps used in the SSCAIT.

Genetic Algorithm Implementation

In this release we implemented a genetic algorithm that determines building placement in our base.
For this release the genetic algorithm has only been implemented, not trained. The placement of the buildings is random at this point and will be trained for the next release.

Post-Tournament Code Cleanup

This release is a cleaner version of the tournament release. The team examined each class and cleaned up the code, making sure that it follows established coding standards and that each segment is written clearly and concisely.

Tournament Release

Version of the agent that has been submitted to the SSCAIT.

The agent utilizes the same design that was implemented in the Alpha release but with enhancements to the strategy and mechanics. The agent can now beat the built-in agent that is provided in the StarCraft Brood War game.

Alpha Release

This release of the bot has the same functionality as our prototype but it has been implemented in the new code structure we designed.

Prototype

This release is the first working version of our bot.

The goals for this release are as follows

• Show that we can successfully connect our bot to a game of StarCraft Brood War
• Show that we can successfully utilize BWAPI commands
• Implement a hard coded game strategy
• Win against a passive* player

*passive player - a player that only builds worker units and gathers resources and nothing else

All of these goals were successfully met by this release