EgoAVFlow: Robot Policy Learning with Active Vision from Human Egocentric Videos via 3D Flow

TL;DR

EgoAVFlow enables robots to learn manipulation and active vision from egocentric videos by predicting actions and viewpoints using 3D flow and diffusion models, improving visibility and manipulation robustness without robot demonstrations.

Contribution

It introduces a novel approach combining 3D flow representations and diffusion models for active vision and manipulation learning from egocentric videos, transferring capabilities without robot demonstrations.

Findings

Outperforms prior human-demo-based baselines in real-world tests

Maintains visibility effectively during manipulation tasks

Robustly adapts to changing viewpoints

Abstract

Egocentric human videos provide a scalable source of manipulation demonstrations; however, deploying them on robots requires active viewpoint control to maintain task-critical visibility, which human viewpoint imitation often fails to provide due to human-specific priors. We propose EgoAVFlow, which learns manipulation and active vision from egocentric videos through a shared 3D flow representation that supports geometric visibility reasoning and transfers without robot demonstrations. EgoAVFlow uses diffusion models to predict robot actions, future 3D flow, and camera trajectories, and refines viewpoints at test time with reward-maximizing denoising under a visibility-aware reward computed from predicted motion and scene geometry. Real-world experiments under actively changing viewpoints show that EgoAVFlow consistently outperforms prior human-demo-based baselines, demonstrating effective visibility maintenance and robust manipulation without robot demonstrations.