A framework for modelling believable behaviour of ambient non-player characters

Abstract

Computer games often feature virtual computer-controlled characters, Non-Player Characters (NPCs). They are an essential part of the game world, providing the player with context, challenges, and interactions. Ambient NPCs are background characters that are used to simply populate the game world and provide limited significance to a game’s story line. However, they play a crucial role in enhancing the player’s immersive experience by creating a sense of realism. To achieve this, ambient NPCs must behave in a way that is perceived as believable by the player. Modelling human behaviour is a challenging task because humans are inherently unpredictable, and their decision-making process is affected by various factors. This project’s primary focus was to develop a framework aimed at modelling believable ambient NPCs to create believable populations in game worlds. The proposed framework draws on previous research investigating the characteristics contributing to believability. In particular, it models goal-based behaviour based on intentionality, motivation and characterhood. The framework is developed as an ECS implementation to allow flexibility and adaptability in implementing varying simulation environments. It uses a fuzzy rule-based system to emulate the unpredictability in human decision-making to generate more realistic behaviour. A case study simulating a university environment was developed to demonstrate the practical application of the framework and evaluate its capabilities in modelling believable behaviour. The behavioural traces generated by the agents in multiple simulation runs were used to systematically analyse the behaviour at individual and population levels. As a result, the functionality and effectiveness of the framework were assessed, revealing its success in modelling realistic goal-driven behaviour. Although there is room for improvement to enhance the diversity and variety in the overall population behaviour, the framework provides a solid foundation for simulating behaviour in diverse simulation environments and shows potential for further development.