• Corpus ID: 220546314

Co-generation of game levels and game-playing agents

@inproceedings{Dharna2020CogenerationOG,
  title={Co-generation of game levels and game-playing agents},
  author={Aaron Dharna and Julian Togelius and Lisa B. Soros},
  booktitle={AIIDE},
  year={2020}
}
Open-endedness, primarily studied in the context of artificial life, is the ability of systems to generate potentially unbounded ontologies of increasing novelty and complexity. Engineering generative systems displaying at least some degree of this ability is a goal with clear applications to procedural content generation in games. The Paired Open-Ended Trailblazer (POET) algorithm, heretofore explored only in a biped walking domain, is a coevolutionary system that simultaneously generates… 

Figures and Tables from this paper

Watts: Infrastructure for Open-Ended Learning
TLDR
Examining implementations of three OEL algorithms, the paper introduces the modules of the Watts framework, a framework motivated by modularity and algorithmic infallibility for implementing, comparing, and recombining open-ended learning (OEL) algorithms.
Transfer Dynamics in Emergent Evolutionary Curricula
TLDR
The main question addressed is how the open-ended learning actually works, focusing in particular on the role of transfer of policies from one evolutionary branch (“species”) to another, and the most insightful finding is that inter-species transfer is crucial to the system’s success.
Coevolution of neural networks for agents and environments
TLDR
The algorithm detailed here demonstrates how artificial neural networks can be evolved in a process of neuroevolution and used as agents in the 2D-game Zelda, producing relevant behaviors.
Deep Surrogate Assisted Generation of Environments
TLDR
Results in two benchmark domains show that DSAGE outperforms existing QD environment generation algorithms in discovering collections of environments that elicit diverse behaviors of a state-of-the-art RL agent and a planning agent.

References

SHOWING 1-10 OF 31 REFERENCES
General Video Game AI: A Multitrack Framework for Evaluating Agents, Games, and Content Generation Algorithms
TLDR
This survey paper presents the VGDL, the GVGAI framework, existing tracks, and reviews the wide use of GVgaI framework in research, education, and competitions five years after its birth.
Fully Differentiable Procedural Content Generation through Generative Playing Networks
TLDR
The capability of Generative Playing Networks, a framework that learns agent policies and generates environments in tandem through a symbiotic process, is demonstrated by training an agent and level generator for a 2D dungeon crawler game.
POET: open-ended coevolution of environments and their optimized solutions
TLDR
The Paired Open-Ended Trailblazer (POET) algorithm is introduced, which generates diverse and sophisticated behaviors that create and solve a wide range of environmental challenges, many of which cannot be solved by direct optimization, or by a direct-path curriculum-building control algorithm.
Grandmaster level in StarCraft II using multi-agent reinforcement learning
TLDR
The agent, AlphaStar, is evaluated, which uses a multi-agent reinforcement learning algorithm and has reached Grandmaster level, ranking among the top 0.2% of human players for the real-time strategy game StarCraft II.
Procedural Content Generation in Games
This book presents the most up-to-date coverage of procedural content generation (PCG) for games, specifically the procedural generation of levels, landscapes, items, rules, quests, or other types of
Investigating vanilla MCTS scaling on the GVG-AI game corpus
  • M. Nelson
  • Computer Science
    2016 IEEE Conference on Computational Intelligence and Games (CIG)
  • 2016
TLDR
This work performs scaling experiments on the potential performance of the baseline vanilla MCTS controller and finds that given enough computational budget, it manages to avoid explicitly losing on most games, despite failing to win them and ultimately losing as time expires, suggesting an asymmetry in the current GVG-AI competition's challenge.
Mastering the game of Go with deep neural networks and tree search
TLDR
Using this search algorithm, the program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0.5, the first time that a computer program has defeated a human professional player in the full-sized game of Go.
Evolutionary Game Design
  • C. Browne
  • Computer Science
    IEEE Trans. Comput. Intell. AI Games
  • 2010
TLDR
The concept of game quality, its automated measurement through self-play simulations, and its use in the evolutionary search for new high-quality games are examined.
An experiment in automatic game design
TLDR
This paper presents a first attempt at evolving the rules for a game, and explains the rationale for doing this and surveys the theories of entertainment and curiosity that underly the fitness function.
Artificial Intelligence and Games
TLDR
This is the first textbook dedicated to explaining how artificial intelligence techniques can be used in and for games, and how to use AI to play games, to generate content for games and to model players.
...
...