Planning in the midst of chaos: how a stochastic Blood Bowl model can help to identify key planning features

For several decades now, games have become an important research ground for artificial intelligence. In addition to often present useful and complex problems, they also provide a clear framework thanks to their rules, sometimes numerous. In this article, we explore a very difficult two-players board game named Blood Bowl. This game allows the players to perform many different actions, which depend for a large part on the result of one or more dice rolls. Thus, it can be seen as a multi-action probabilistic problem driven by a Markov decision process. In this article, we present the first stochastic model of the main phase of Blood Bowl to our knowledge and the premise of a dedicated planning framework. Such a framework could offer interesting grounds and insights for modeling high turn-wise branch factor games.

[1]  Peter I. Cowling,et al.  Evolutionary MCTS for Multi-Action Adversarial Games , 2018, 2018 IEEE Conference on Computational Intelligence and Games (CIG).

[2]  Csaba Szepesvári,et al.  Bandit Based Monte-Carlo Planning , 2006, ECML.

[3]  Julian Togelius,et al.  Blood Bowl: A New Board Game Challenge and Competition for AI , 2019, 2019 IEEE Conference on Games (CoG).

[4]  R. Bellman A Markovian Decision Process , 1957 .

[5]  Leslie Pack Kaelbling,et al.  On the Complexity of Solving Markov Decision Problems , 1995, UAI.

[6]  Julian Togelius,et al.  Artificial Intelligence and Games , 2018, Springer International Publishing.

[7]  Yishay Mansour,et al.  A Sparse Sampling Algorithm for Near-Optimal Planning in Large Markov Decision Processes , 1999, Machine Learning.

[8]  Sabarinath Mohandas,et al.  A.I for Games with High Branching Factor , 2018, 2018 International CET Conference on Control, Communication, and Computing (IC4).

[9]  Shivakumar Vaithyanathan,et al.  Generalized Model Selection for Unsupervised Learning in High Dimensions , 1999, NIPS.

[10]  Andreas Gustafsson Winner Prediction of Blood Bowl 2 Matches with Binary Classification , 2019 .

[11]  Julian Togelius,et al.  Online Evolution for Multi-action Adversarial Games , 2016, EvoApplications.

[12]  Janet Wiles,et al.  The challenge of Go as a domain for AI research: a comparison between Go and chess , 1995, Proceedings of Third Australian and New Zealand Conference on Intelligent Information Systems. ANZIIS-95.