COMBAT: Conditional World Models for Behavioral Agent Training

2026-02-28

AI Summary

This work addresses the challenge that existing world models struggle to generate dynamic opponents capable of intelligent responses to player behavior when trained solely on single-player data. The authors propose the first diffusion-based, real-time, action-controllable world model, leveraging a 1.2-billion-parameter diffusion Transformer architecture integrated with a deeply compressed autoencoder, causal distillation, and diffusion guidance. Without explicit policy supervision, the model implicitly learns and exhibits emergent interactive behaviors. It supports real-time inference and effective learning under partial observability, successfully generating responsive and strategic opponents in *Tekken 3*. To validate the emergence of intelligent agent behavior, the study also introduces a novel evaluation metric.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Multiagent Systems: Adversarial Agents
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Recent advances in video generation have spurred the development of world models capable of simulating 3D-consistent environments and interactions with static objects. However, a significant limitation remains in their ability to model dynamic, reactive agents that can intelligently influence and interact with the world. To address this gap, we introduce COMBAT, a real-time, action-controlled world model trained on the complex 1v1 fighting game Tekken 3. Our work demonstrates that diffusion models can successfully simulate a dynamic opponent that reacts to player actions, learning its behavior implicitly. Our approach utilizes a 1.2 billion parameter Diffusion Transformer, conditioned on latent representations from a deep compression autoencoder. We employ state-of-the-art techniques, including causal distillation and diffusion forcing, to achieve real-time inference. Crucially, we observe the emergence of sophisticated agent behavior by training the model solely on single-player inputs, without any explicit supervision for the opponent's policy. Unlike traditional imitation learning methods, which require complete action labels, COMBAT learns effectively from partially observed data to generate responsive behaviors for a controllable Player 1. We present an extensive study and introduce novel evaluation methods to benchmark this emergent agent behavior, establishing a strong foundation for training interactive agents within diffusion-based world models.

Problem

• world models
• dynamic agents
• reactive behavior
• interactive simulation
• behavioral modeling

Innovation

• conditional world models
• diffusion transformer
• emergent agent behavior
• real-time inference
• behavioral agent training