CI in Games

Solving Problems in Games
by Means of
Computational Intelligence

Current Activities

Members

Publications

Journal Articles

  • Daniele Loiacono, Pier Luca Lanzi, Julian Togelius, Enrique Onieva, David A. Pelta, Martin V. Butz, Thies D. Lönneker, Luigi Cardamone, Diego Perez, Yago Saez, Mike Preuss, Jan Quadflieg. The 2009 Simulated Car Racing Championship. IEEE Transactions on Computational Intelligence and AI in Games, Volume 2, Issue 2, pp. 131-147. 2010.

Conference Articles (peer reviewed)

  • Philip Hingston, Mike Preuss, and Daniel Spierling. RedTNet: A Network Model for Strategy Games. Proc. of the 2010 IEEE Conference on Evolutionary Computation (CEC), part of 2010 IEEE World Congress on Computational Intelligence (WCCI 2010), IEEE Press, 2010.

Technical Reports

  • Holger Danielsiek, Christian Eichhorn, Tobias Hein, Edina Kurtić, Georg Neugebauer, Nico Piatkowski, Jan Quadflieg, Sebastian Schnelker, Raphael Stüer, Andreas Thom, Simon Wessing. PG511 - CI in Games - Final Report. Reihe CI 252/08, SFB 531, Technische Universität Dortmund, 2008.
  • Holger Danielsiek, Christian Eichhorn, Tobias Hein, Edina Kurtić, Georg Neugebauer, Nico Piatkowski, Michael Puchowezki, Jan Quadflieg, Sebastian Schnelker, Raphael Stüer, Andreas Thom, Simon Wessing. PG511 - CI in Games - Zwischenbericht. Reihe CI 236/07, SFB 531, Technische Universität Dortmund, 2007.

Demonstration Articles

  • Niels Ackermann, Alireza Gholaman, Matthias Grochowski, Marc Gorzala, Markus Kemmerling, Daniel Spierling, Thomas Harweg, Sebastian Uellenbeck, Wolfgang Walz. Emotionally affected poker players. Demonstration Paper. Workshop within 10th International Conference on Parallel Problem Solving From Nature (PPSN 2008), 2008.
  • Nicola Beume, Boris Naujoks, Georg Neugebauer, Nico Piatkowski, Mike Preuss. CI enlivened ghosts for Pac-Man. Demonstration Paper. Workshop within 10th International Conference on Parallel Problem Solving From Nature (PPSN 2008), 2008.

Teaching

Completed Project Groups

  • PG 511: CI in games - Methoden der Computational Intelligence zur Entwicklung von Spielstrategien. SoSe 2007 - WiSe 2007/08
    Advisors: Beume, Naujoks, Preuß.
  • PG 529: Spielercharaktere mit CI - Modellierung menschenähnlicher Gegenspieler in Strategiespielen mit Techniken der Computational Intelligence. SoSe 2008 - WiSe 2008/09
    Advisors: Beume, Naujoks, Preuß.
  • PG 554: Mobile Musik-Welten – Spielerische Exploration von digitalen Musiksammlungen. SoSe 2011 - WiSe 2011/12
    Advisors: Vatolkin, Preuß.

Current Project Group

None

Current Diploma Theses

None

Completed Diploma Theses

  • Markus Kemmerling: Optimierung der Fahrzeugabstimmung auf Basis verrauschter Daten einer Autorennsimulation.
    Advisors: Rudolph, Preuß.
  • Niels Ackermann: Evolutionäre Spielbaumsuche für das Strategiespiel Diplomacy.
    Advisors: Rudolph, Preuß.
  • Jan Quadflieg: Lernende und planende Steuerungen für eine Autorennsimulation. February 2010.
    Advisors: Rudolph, Preuß.
  • Daniel Haus: Ein Konzept zur Unterstützung menschlicher Spieler in Browsergames. July 2009.
    Advisors: Rudolph, Preuß.
  • René Goebels: Methoden der Computational Intelligence zur Gegnermodellierung beim Pokerspiel. October 2008.
    Advisors: Rudolph, Beume.
  • André Briskorn: Neuronales Selbstlernen von strategischen Spielen. September 2006.
    Advisors: Rudolph, Jansen.

Current Master Theses

None

Completed Master Theses

  • Matthias Kuchem: Reaktive Pareto-Optimale Eröffnungsstrategien in Echtzeitstrategiespielen. April 2014.
    Advisors: Rudolph, Preuß.

Current Bachelor Theses

  • Kevin Majchrzak: Adaptives Scripting für Echtzeitspiele.
    Advisors: Rudolph, Quadflieg.

Completed Bachelor Theses

  • Fabian Bruckner: Robuste Rennlinienplanung in der Rennsimulation TORCS.
    Advisors: Rudolph, Quadflieg.
  • Jan Mühlig: Kontrollierbare Schwarmbewegung von Einheiten in Starcraft.
    Advisors: Rudolph, Quadflieg.
  • Janine Hemmers: Prozedurale Generierung von Spielfeldern für ein Tower Defense Duell.
    Advisors: Rudolph, Quadflieg.
  • Kai Verlage: Adaptive Ausweichstrategien für die Autorennsimulation TORCS.
    Advisors: Rudolph, Quadflieg.
  • Ersin Özdemir: Computergegner für das Strategiespiel “Triple Triad” auf mobilen Endgeräten.
    Advisors: Rudolph, Quadflieg.
  • Micheal Pantförder: Realisierung eines Jump 'n' Run Spiels unter Verwendung der Unity3D Game-Engine und des KINECT Sensors. February 2013.
    Advisors: Rudolph, Quadflieg.
  • Marco Kuhnke: Konzeption eines Multiplayer-Ballspiels auf Basis der Game Engine Unity3D. December 2012.
    Advisors: Rudolph, Quadflieg.
  • Christopher Morris: Ameisenalgorithmen zur Lösung des Logikrätsels Inshinoheya. December 2011.
    Advisors: Rudolph, Quadflieg.
  • Tim Delbrügger: Modellierung des Lenkverhaltens eines Fahrzeugs in einer Autorennsimulation. September 2010.
    Advisors: Rudolph, Preuß.
  • Susanna Pohl: Kollisionsvermeidung in einer Autorennsimulation. October 2010.
    Advisors: Rudolph, Preuß.
  • Jan Czogalla: Ein reaktiver Computergegner für ein Echtzeitstrategiespiel.
    Advisors: Rudolph, Preuß.

Completed Seminar

  • Computational Intelligence bei Computerspielen. SoSe 2007.
    Advisor: Rudolph

Project Groups

PG 511

Due to the fact that the Artificial Intelligence (AI) of non-player characters (NPCs) in computer games often features shortfalls, the PG511 tried to improve NPCs' intelligence by means of Computational Intelligence (CI), such as Artificial Neural Nets (ANN) and Evolutionary Algorithms (EA). The hope was that this approach would make their behaviour less predictable. The main task of the PG511 was to figure out whether CI methods could be applied to computer games, how far CI methods can balance a game's difficulty while giving the observer an impression that the NPCs behave in a natural, human-like way. As mentioned above the PG511 had two semesters for developing CI methods for computer games, a Pac-Man clone called NJam and a more complex realtime strategy game (RTS) Glest. For further information you can visit the PG511 homepage .

PG 529

The aim of the PG529 was to enhance gameplay experience by modeling more humanlike non-player characters. During the two semester period the PG529 worked out solutions for two very different games. The first game was the computer game PokerTH which is a computer version of the popular poker variant Texas Hold'em. In the process of enhancing the game, the project group removed the existing computer opponent and replaced it by an opponent being controlled by an Artificial Neural Net (ANN). The ANN was optimized using an Evolutionary Algorithm (EA) and was able to reproduce the behaviour of human players. Furthermore two new ways of communication were added: Sending predefined text messages and the ability of showing emoticons. The second game was the computer version of the board game Diplomacy. Negotiation is a very important aspect while playing Diplomacy which is why the main target of the PG529 was to improve the non-player characters ability to negotiate and evaluate the relationships between the different players. The group was divided into three sub-groups which developed different bots implementing different strategies independently. The developed bots where compared and a Turing test has been performed.

Stragotiator

One of the bots that were first developed within the project group have been taken further and is also available from the web page of Markus Kemmerling. A description of the major features and the planning method that has been added by Niels Ackermann is contained in the EvoGames 2011 paper (EvoApplications 2011: Proceedings, Part I, Springer).

Fachprojekte (technical projects)

DET SS 2012

Course plan (in German), linking to some additional material

IBalls game the 3rd of 3 projects, a not-so-trivial multiplayer game for the Unity enginge, see example picture (at startup time). It was developed with help of modern design tools as articy:draft.

 
Last modified: 2014-09-13 23:52 by Günter Rudolph
DokuWikiRSS-Feed