Griddly: A platform for AI research in games

TL;DR

Griddly is a versatile platform that enables researchers to easily prototype and test reinforcement learning agents across various configurable grid-based games, facilitating studies on generalization and overfitting.

Contribution

The paper introduces Griddly, a new platform combining configurable game environments, diverse observer types, and an efficient core engine to advance AI research in games.

Findings

RL agents' performance varies with different observation configurations.

Platform supports rapid prototyping of diverse game environments.

Baseline experiments highlight factors affecting RL generalization.

Abstract

In recent years, there have been immense breakthroughs in Game AI research, particularly with Reinforcement Learning (RL). Despite their success, the underlying games are usually implemented with their own preset environments and game mechanics, thus making it difficult for researchers to prototype different game environments. However, testing the RL agents against a variety of game environments is critical for recent effort to study generalization in RL and avoid the problem of overfitting that may otherwise occur. In this paper, we present Griddly as a new platform for Game AI research that provides a unique combination of highly configurable games, different observer types and an efficient C++ core engine. Additionally, we present a series of baseline experiments to study the effect of different observation configurations and generalization ability of RL agents.