Search

In this study, we aim to explore the characteristics of different AI algorithms using the classical Chinese board game xiangqi as the testbed. We did this by utilizing four different build-in AI agents in Ludii: Random, Flat MC, UCT and Minimax, and having them playing against each other in all possible match ups and exploring their play patterns. The result shows that in terms of the most basic play patterns, i.e., the win/loss rate, we can arrive at a skill ranking as follows: Minimax > UCT > Flat MC > Random. When it comes to more specific metrics such as piece frequency and taking out, Minimax also stands out from all the other three AI agents. We believe this study is a good starting point for looking into deeper how these AI agents differ from each other. The study also suggests that just the type of algorithm and how it is implemented can already possibly affect the observed play patterns and for further research exploring mapping between AI and play types it is important to be familiar with the default playing pattern of these AI algorithms before turning them into specific player types.

Game boards are described in the Ludii general game system by their underlying graphs, based on tiling, shape and graph operators, with the automatic detection of important properties such as topological relationships between graph elements, directions and radial step sequences. This approach allows most conceivable game boards to be described simply and succinctly.

This paper proposes Heuristic Sampling (HS) for generating self-play trials for games with a defined state evaluation function, with speeds comparable to random playouts but game length estimates comparable to those produced by intelligent AI agents. HS produces plausible results up to thousands of times faster than more rigorous methods.

Many games often share common ideas or aspects between them, such as their rules, controls, or playing area. However, in the context of General Game Playing (GGP) for board games, this area remains under-explored. We propose to formalise the notion of "game concept", inspired by terms generally used by game players and designers. Through the Ludii General Game System, we describe concepts for several levels of abstraction, such as the game itself, the moves played, or the states reached. This new GGP feature associated with the ludeme representation of games opens many new lines of research. The creation of a hyper-agent selector, the transfer of AI learning between games, or explaining AI techniques using game terms, can all be facilitated by the use of game concepts. Other applications which can benefit from game concepts are also discussed, such as the generation of plausible reconstructed rules for incomplete ancient games, or the implementation of a board game recommender system.

The Digital Ludeme Project (DLP) aims to recon-struct and analyse over 1000 traditional strategy games using modern techniques. One of the key aspects of this project is the development of Ludii, a general game system that will be able to model and play the complete range of games required by this project. Such an undertaking will create a wide range of possibilities for new AI challenges. In this paper we describe many of the features of Ludii that can be used. This includes designing and modifying games using the Ludii game description language, creating agents capable of playing these games, and several advantages the system has over prior general game software

Ludii is a general game system being developed as part of the ERC-funded Digital Ludeme Project (DLP). While its primary aim is to model, play, and analyse the full range of traditional strategy games, Ludii also has the potential to support a wide range of AI research topics and competitions. This paper describes some of the future competitions and challenges that we intend to run using the Ludii system, highlighting some of its most important aspects that can potentially lead to many algorithm improvements and new avenues of research. We compare and contrast our proposed competition motivations,goals and frameworks against those of existing general game playing competitions, addressing the strengths and weaknesses of each platform.

While there exist a variety of game description languages (GDLs) for modelling various classes of games, these are aimed at game playing rather than the more particular needs of game design. This paper describes a new approach to general game modelling that arose from this need. A class grammar is automatically generated from a given library of source code, from the constructors and associated parameters found along its class hierarchy, to give a context-free grammar that provides access to the underlying code while hiding its implementation details.