AI for General Strategy Game Playing

Jon Lau Nielsen, Benjamin Fedder Jensen, Tobias Mahlmann, Julian Togelius, Georgios N. Yannakakis

    Research output: Chapter in Book/Report/Conference proceedingChapter


    This chapter addresses the understudied question of how to create AI that plays strategy games, through building and comparing AI for general strategy game playing. The chapter enumerates a series of common characteristics of such a game. Many strategy games can be played in multiplayer mode, where human players compete with each other for domination. There has been extensive research done on AI for traditional board games. The chapter addresses the problem of general strategy game playing. The Strategy Game Description Language (SGDL) is a model-based approach to develop strategy games. The chapter presents 11 agents that are created based on several different techniques. The MinMax, Monte Carlo tree search (MCTS), potential field (PF), and neuroevolution of augmenting topologies (NEAT) agents were determined to be adequate in various models and map combinations and thus are capable of general game play.

    Original languageEnglish (US)
    Title of host publicationHandbook of Digital Games
    PublisherWiley-IEEE Press
    Number of pages31
    ISBN (Electronic)9781118796443
    ISBN (Print)9781118328033
    StatePublished - Apr 7 2014


    • AI
    • Board games
    • MinMax
    • Monte Carlo tree search (MCTS)
    • Neuroevolution of augmenting topologies (NEAT) agents
    • Potential field (PF)
    • Strategy game
    • Strategy game descriptionlanguage (SGDL)

    ASJC Scopus subject areas

    • General Engineering


    Dive into the research topics of 'AI for General Strategy Game Playing'. Together they form a unique fingerprint.

    Cite this