Skip to content

shriuken/starcraft2ai

Repository files navigation

Starcraft 2 AI

This document is adapted, in part, from the original project description document.

Table of Contents

  1. Project Abstract
    1. Team Members
    2. Advisor
  2. Project Description
  3. User Stories and Design Diagrams
    1. User Stories
    2. Design Diagrams
      1. Level 0
      2. Level 1
      3. Level 2
  4. Project Tasks and Timeline
  5. ABET Concerns Essay
  6. Slideshow
    1. Slides
    2. Recorded Presentation
  7. Self-Assessment Essays
  8. Professional Biographies
  9. Budget
  10. Appendix
  11. TODO

Abstract

There have been many recent advances in game-playing AIs, such as the Dota2 AI and AlphaGo. With this project, we explored the use of conventional and cutting edge ML techniques to create a self-learning Starcraft II (SC2) AI agent capable of defeating Blizzard's Very Easy AI.

Team Members

Kyle Arens - [email protected]
Ryan Benner - [email protected]
Jon Deibel - [email protected]

Faculty Advisor

Dr. Ali Minai - http://www.ece.uc.edu/~aminai/

Project Description

Inspired by recent advances in game-playing AI, we are attempting to address a similarly themed, albeit different problem - creating an SC2 AI agent that operates at human APM (actions per minute), up to ~400apm. Current SC2 AI agents abuse the fact that agents are able to take an inhuman number of actions extraordinarily quickly (such as 5000apm). Our initial goal was to restrict APM down to ~300apm in order to model our AI off of human capabilities and to more-effectively learn decisions. However, we settled on an upper limit of ~480apm, which is still human attainable by the most skilled players. With a hard-limit of ~480apm, any given action taken will have more importance than if the bot had a higher limit, such as ~5000apm.

We are scoping our project in the following manner:

  • the agent will play as Terran
  • against a Very Easy Terran AI provided in the retail version of the game
  • on the maps Simple64 and flat64.

User Stories & Design Diagrams

User Stories

User Stories

Design Diagrams

Level 0

D0 diagram

This diagram represents the highest level view of our architecture. Starcraft 2 game replay data is taken in, and used to generate a model for autonomously playing the game.

Level 1

D1 diagram

This diagram elaborates on our AI engine, and gives a high level view of what is being done at each stage. First the game state is parsed, then that parsed data is used to train the model, and turn the model decisions into in-game actions.

Level 2

D2 diagram

This diagram gives a much more in depth breakdown of what each portion of our AI engine does. It elaborates greatly on how the data is parsed, and how it's used in long and short term planning by our model.

Project Tasks and Timelines

Combined Task List, Timeline, and Effort Matrix.

ABET Concerns Essays

Essay.

Slideshow

Slides

Fall Presentation. Spring Presentation.

Recorded Presentation

Fall Video. A recorded presentation was not created for Spring Semester.

Each member wrote their own assesment essay.

Budget

There have been no expenses to date.

Appendix

References

Python abstraction: https://github.com/deepmind/pysc2
Raw C++ API for starcraft: https://github.com/Blizzard/s2client-proto
For deep model creation: https://github.com/tensorflow/tensorflow
Silver, David, et al. "Mastering the game of go without human knowledge." Nature 550.7676 (2017): 354-359.

Work breakdown

Work Time Team Members
Learning the basics of Reinforcement Learning 25 Hours Kyle Arens, Jon Deibel, Ryan Benner
Understanding C++/Python API 10 Hours Jon Deibel, Ryan Benner, Kyle Arens
Creation of baseline AI/First RL based AI 10 Hours Jon Deibel, Ryan Benner
Documentation 5 Hours Ryan Benner, Kyle Arens, Jon Deibel
Reading + Understanding AlphaGo 5 Hours Ryan Benner, Kyle Arens, Jon Deibel

TODO

  • Update design diagrams.
  • Update task list (further).
  • Research Actor-Critic Model.
  • Continue research on Q-tables and Q-functions.
  • Discussions on game state.

About

Keep it one line.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •  

Languages