Home

Note: This site is auto-converted from the classic version, which is forced to be unavailable, please expect issues including missing images, missing links, formatting and layout issues etc.


This project centers around exploring the power of structured procedural content generation in game design and experimenting machine learning techniques on top of the complex structures produced by the procedural generator. We create a prototype card game (called Chaos Cards) similar to Hearthstone (with some modification to simplify or accommodate to the need of our research). A procedural card generator is implemented in the game, with GIGL. A texted version simulator and a pure random AI player are also implemented, in C++. This project is currently under progress, the machine learning part is planned but not implemented yet.

The above shows some demos about Chaos Cards. The left part shows an example fragment of console text visual of a match. ``..." indicates omission and blue text indicates console input. The right side shows example cards procedurally generated. All card names are filled in with their generation random seed. All examples here are produced with no machine learning or any other balancing schemes.

Programming Language:

GIGL (a DSL I developed for grammatical PCG) and C++.

Status:

Inactive.

The Developer:

Tiannan "Fenix" Chen

The Advisor:

Prof. Stephen J. Guy