Abstract
In most games, the decision-making of non-player characters (NPCs) is usually constructed using variants of state machines, behaviour trees, utility-based AI or planning. These methods are relatively simple to implement, but have drawbacks in that it can be difficult to create complex non-hardcoded behaviour for many agents and to maintain the algorithms, especially when scaling up. Game designers usually think of their games with rules that closely resemble logic rules. A methodology is introduced to design both general and modular behaviour using a logic reasoner with hierarchical ontologies. This approach is combined with the well-founded semantics (WFS) to solve the problem of representation and reasoning despite the lack of NPC knowledge.
Sylvain LAPEYRADE
Data Scientist & PhD Student
My interests include Artificial Intelligence, Data Science, Machine Learning and Games.