Emergent Narrative
Exploring the true potential of storytelling in games by letting the story emerge from the player’s interaction with the virtual world.
Unlike traditional media such as books and movies, games offer a unique opportunity for storytelling: true interactivity. However, many video game titles tell linear stories on which the player has little to no impact. By contrast, we are exploring how stories can emerge by simulating worlds and their inhabitants. But there are open questions: How can designers affect the stories? How can we make sure that stories will be interesting? Can games run these simulations in real-time? By finding answers to these challenges we hope to unlock the true potential of storytelling in interactive media.
NPC engine: MCTS Planning for Emergent Narrative
In video games, the developers' desire to provide an engaging story is often at odds with the player's wish to influence the story. We believe that emergent narrative can resolve this conflict. In emergent narrative stories aren't scripted – instead, they result from the simulation of characters in a story world. Our NPC engine achieves complex character behavior by having each character plan individually, combining simple behavior definitions and Monte Carlo Tree Search (MCTS). Characters consider other nearby characters, leading to complex social behavior. With self learning, our characters can also improve their behavior over time.
Publications
Student Projects
Showcasing Complex Behaviour Emergence in Multi-agent Simulation
by Heinrich Grattenthaler
Learning Action Value Functions for Accelerating Multi-agent Planning for Emergent Narrative
by Nora Tommila
Learning State Value Functions for Accelerating Multi-agent Planning for Emergent Narrative
by Patrick Eppensteiner
Emergent Narrative through Reasoning Agents in Location-Based Multiplayer Games
by Henry Raymond