Highlights:
- Introduces Regular Games (RG), a new automata-based framework for general game playing (GGP).
- Combines computational efficiency with a flexible design language for human and procedural game creation.
- Outperforms existing GGP frameworks such as Regular Boardgames and Ludii in speed and efficiency.
- Includes an advanced ecosystem for visualization, debugging, and benchmarking.
TLDR:
Researchers from Poland unveil Regular Games, an automata-based General Game Playing system that enhances efficiency and ease of design, surpassing current frameworks and offering a robust suite of tools for AI-driven game creation and analysis.
A team of researchers has introduced **Regular Games (RG)**, a breakthrough framework poised to redefine the field of **General Game Playing (GGP)** within artificial intelligence. Developed by **Radosław Miernik**, **Marek Szykuła**, **Jakub Kowalski**, **Jakub Cieśluk**, **Łukasz Galas**, and **Wojciech Pawlik**, this system addresses a central challenge in GGP: designing a language that is both computationally efficient and convenient for human and automated game creation.
At the heart of Regular Games lies a **low-level language** based on **finite automata**, a mathematical structure adept at representing and processing discrete game states and transitions. This minimalistic design enables agents and algorithms to efficiently interpret, analyze, and optimize game models while maintaining universal expressiveness for **all finite turn-based games with imperfect information**. The architecture’s simplicity not only accelerates computation but also lends itself to procedural content generation and automated reasoning, bridging the gap between AI theory and practical game design workflows.
Beyond its core, the Regular Games framework supports a hierarchy of higher-level languages that are directly translatable into the low-level automata representation. This layered approach ensures that human designers—or AI-driven procedural systems—can create complex, multi-agent games in a more intuitive way without compromising performance. The authors report that RG’s **forward model generation** outperforms leading systems such as **Regular Boardgames** and **Ludii**, providing substantial gains in speed and versatility. Crucially, the RG ecosystem also includes a comprehensive set of development tools: an **LSP-integrated editor**, **automaton visualization interface**, **benchmarking utilities**, and a **debugger for game description transformations**. Together, these enhancements mark a substantial step forward in building scalable, transparent, and AI-friendly platforms for game research and development.
This innovation represents a milestone for AI researchers and developers exploring autonomous strategy generation, multi-agent simulations, and creative AI systems. By merging the rigor of automata theory with modern computational tools, Regular Games provides a foundation that could streamline not only GGP experimentation but also broader applications in reinforcement learning and digital content automation.
Source:
Source:
Original Research Paper: “Regular Games — an Automata-Based General Game Playing Language” by Radosław Miernik, Marek Szykuła, Jakub Kowalski, Jakub Cieśluk, Łukasz Galas, and Wojciech Pawlik, arXiv:2511.10593 [cs.AI], https://doi.org/10.48550/arXiv.2511.10593