Competition used


More about module


Prof. Julian Togelius

New York University, USA

Ms. Pac-Man

Artificial Intelligence

(introductory level)

A semester-long project where students implemented the following algorithms within this framework:

DFS, BFS, Iterative deepening, A*, Hill-climber, Simulated annealing, Genetic algorithm, Evolution strategy, k-Nearest neighbor, Perceptron, ID3 decision tree, Q-learning.

Plan to use the same frameworks similarly next year.

Course slides


Artificial Intelligence in Games

(advanced level)

Prof. Gillian Smith

Northeastern University, UK



Mario level generation framework

Game AI

(cross-listed undergraduate and master's level)

Students use the PacMan framework to implement the original (or as close to it as the framework allows...) ghost AI algorithm and then try their hand at making a better one.

Students use the mario framework to write a human-like controller for Mario, and we do our own internal bracket-based competition that's held independent from grading considerations.

Dr. José María Font Fernández

Universidad Politécnica de Madrid, Spain

Ms. Pacman

Machine Learning (undergrad)

Students use Ms. Pacman to implement decision trees and neural networks that learned from their own gameplay. These systems were used for the Pacman AI.


Evolutionary Computation (undergrad)

Students coded a genetic algorithm to evolve an intelligent system of their choice (fuzzy-rule based systems and neural networks were the most popular), for the Ghosts AI.

Students also coded a grammar-guided genetic program to evolve levels for the Mario framework. Evaluation was completely static: the levels were parsed and analysed to get a fitness score (no AI system played the levels as Mario to get a more detailed evaluation).

Dr. Diego Perez

University of Essex, UK



Game Artificial Intelligence

Prof. Simon Lucas

Dr. Diego Perez

Dr. Jialin Liu

University of Essex, UK



Game Design

(phd and master level)

Two weeks’ full day training. Students use both frameworks to tune game parameters and design agents using evolutionary algorithms or MCTS variants.

Module description, structure and materials in 2017:

Dr. Antonio J. Fernández Leiva

Universidad de Málaga, Spain

Ms. Pacman competition framework, the GVGAI software, and also the UT2004

MsC thesis

Students are using these frameworks to carry out a basic research on the application of AI to videogames.

A multiplayer RTS game is in development to run a future competition.

Ms. Pacman

Basic Game AI (Master in Videogames : Design, Art and Programming)

Students use this software to code the basic techniques used for game AI such as finite states machines, pursuit-evade mechanisms, follow-the-leader techniques, some steering

Dr. Gonzalo Aranda

Universidad de Huelva, Spain



(advanced courses)

In preparation of Machine Learning classes based on Pac Man framework next year.

TORCS cars racing game


(basic courses)

Otto-von-Guericke-University Magdeburg (OVGU), Germany

Fighting Game AI Competition

Computational Intelligence in Games

(Undergraduate and Master's)

In this academic year, they conducted a competition using the fighting game AI platform at the end of their course. 

Sejong University, South Korea

Fighting Game AI Competition

Artificial Intelligence


The platform is used for a first course project, where most of developed AI are rule-bases. For a more-advanced project later, other games are added, and students have to apply CI techniques (reinforcement learning, ensemble algorithm, and so on) to the game of their choice

Prof. Ruck Thawonmas

Ritsumeikan University, Japan

Fighting Game AI Competition

Advanced Topics in Entertainment Computing


Academic Years: Fall Semester 2013, 2014, 2015, 2016, ...

Students were asked to develop their AIs based on techniques like



dynamic scripting

This fall, a sample AI using MCTS + top entries this year will be added to the above list.

For more details on the platform, please check

Prof. Risto Miikkulainen

University of Texas at Austin

NERO game (, and

AI and Neural Networks

(both grad and undergrad)

Students use evolutionary computation and/or reinforcement learning techniques to train a team for the game. We provide a number of teams to test against while they are working on it. After they turn in their solutions, we run a round robin offline, and then semifinals and the final game live in class.

It has been a lot of fun, and highly motivating to the students. Interestingly, the tournaments are also a source of data we can use to analyze both how well the game is working, and the AI techniques in it (see e.g. our CIG paper last year:

Prof. Georgios N. Yannakakis

University of Southern Denmark

various game competitions


various game competitions

University of Mala

various game competitions

Game AI Revisited

(PG level)


Game AI

(PG level)

Ms-Pacman competition was used as a mid-term assignment and a number of competitions were used as final projects.

In 2015, students could pick from

- Platformer AI competition (Turing track)

- Fighting Game AI Competition

- Robocode competition

- The Arcade Learning Environment

- Spelunky

Details about the class/slides/projects/etc can be found here:

Prof. Sanaz Mostaghim

University of Magdeburg

Fighting game competition

Computational Intelligence in Games (BSc and MSc level)