AI-based world behaviour for emergent narratives

• Authors: Jean-Luc Lugrin;Marc Cavazza
• Venue: Proceedings of the 2006 ACM SIGCHI international conference on Advances in computer entertainment technology
• Year: 2006

Research in Interactive Narrative has developed new approaches to the behaviour of virtual actors, but has dedicated little attention to the physical behaviour of the environment in which the action takes place. In this paper, we describe a method supporting the AI-based simulation of object behaviour, so that interactive narrative can feature the physical environment inhabited by the player character as an "actor". The prototype we describe has been developed on top of the Unreal Tournament game engine and relies on a "causal engine", which essentially bypasses the native Physics engine to generate alternative consequences to player interventions. It operates using a small depth-bound planning system which determines the most appropriate object behaviours following player interaction. The prototype is illustrated through a test application called "Death Kitchen", freely inspired from various thriller and horror films, in which the kitchen is plotting against the player character to generate domestic accidents.