ExoActor: Exocentric Video Generation as Generalizable Interactive Humanoid Control

2026-04-30

AI Summary

This work addresses the challenge of generating fluent, interaction-rich behaviors for humanoid robots in complex environments, which requires joint modeling of spatial context, temporal dynamics, and task intent. The authors propose a novel paradigm that leverages third-person video generation as a universal interface for humanoid control. Specifically, a large-scale video generation model synthesizes coherent interactive videos conditioned on task instructions and scene context; these videos are then translated into executable motion sequences via human motion estimation and a general-purpose motion controller. The resulting end-to-end system generalizes to novel scenarios without requiring additional real-world data, demonstrating both the feasibility and scalability of generating diverse, task-oriented humanoid interaction behaviors.

Abstract

Humanoid control systems have made significant progress in recent years, yet modeling fluent interaction-rich behavior between a robot, its surrounding environment, and task-relevant objects remains a fundamental challenge. This difficulty arises from the need to jointly capture spatial context, temporal dynamics, robot actions, and task intent at scale, which is a poor match to conventional supervision. We propose ExoActor, a novel framework that leverages the generalization capabilities of large-scale video generation models to address this problem. The key insight in ExoActor is to use third-person video generation as a unified interface for modeling interaction dynamics. Given a task instruction and scene context, ExoActor synthesizes plausible execution processes that implicitly encode coordinated interactions between robot, environment, and objects. Such video output is then transformed into executable humanoid behaviors through a pipeline that estimates human motion and executes it via a general motion controller, yielding a task-conditioned behavior sequence. To validate the proposed framework, we implement it as an end-to-end system and demonstrate its generalization to new scenarios without additional real-world data collection. Furthermore, we conclude by discussing limitations of the current implementation and outlining promising directions for future research, illustrating how ExoActor provides a scalable approach to modeling interaction-rich humanoid behaviors, potentially opening a new avenue for generative models to advance general-purpose humanoid intelligence.

Problem

• humanoid control
• interaction-rich behavior
• exocentric video generation
• task-conditioned behavior
• generalizable control

Innovation

• exocentric video generation
• humanoid control
• interaction modeling
• task-conditioned behavior
• motion synthesis