EvolvingBehavior: Towards Co-Creative Evolution of Behavior Trees for Game NPCs

TL;DR

EvolvingBehavior is a new tool that uses genetic programming to automatically evolve behavior trees for game NPCs within Unreal Engine 4, aiming to assist developers in creating more goal-oriented AI behaviors.

Contribution

The paper introduces EvolvingBehavior, a novel co-creative tool for evolving behavior trees, demonstrating its ability to produce NPC behaviors comparable to designer-crafted trees.

Findings

EvolvingBehavior can generate behavior trees approaching designer goals.

Evolved behaviors outperform random trees in a 3D survival game.

Discussion on future directions for co-creative AI design tools.

Abstract

To assist game developers in crafting game NPCs, we present EvolvingBehavior, a novel tool for genetic programming to evolve behavior trees in Unreal Engine 4. In an initial evaluation, we compare evolved behavior to hand-crafted trees designed by our researchers, and to randomly-grown trees, in a 3D survival game. We find that EvolvingBehavior is capable of producing behavior approaching the designer's goals in this context. Finally, we discuss implications and future avenues of exploration for co-creative game AI design tools, as well as challenges and difficulties in behavior tree evolution.