Crafting Interactive Narrative Games with Adversarial Planning Agents from Simulations

Abstract

This paper describes Adversario, an implemented system for creating interactive narrative games centered around an adversarial planner directing autonomous agents to achieve goals that conflict with player goals. Through our implementation, we present a novel framework for creating interactive narrative games wherein the autonomous agents guided by the adversarial planner have rich and varied attributes derived from a complex social simulation. By doing this, we show that it is possible to craft interactive narrative games with a set of carefully curated domain actions to produce interactions in a procedurally generated world with rich characters. Our system bridges several gaps between interactive narrative game design, procedural content generation, and classical AI planning. We demonstrate this by transforming a simulation-planning based narrative generation system into a system for streamlining the process of crafting interactive narrative games involving rich procedurally generated agents.