Bridging the Embodiment Gap: Disentangled Cross-Embodiment Video Editing

TL;DR

This paper introduces a generative framework that learns disentangled representations for cross-embodiment video editing, enabling the synthesis of robot demonstrations from human videos without paired data.

Contribution

It proposes a dual contrastive objective to learn independent task and embodiment representations, improving cross-embodiment video synthesis for robotics.

Findings

Produces temporally consistent robot videos from human demonstrations.

Learns disentangled representations without paired cross-embodiment data.

Enables scalable robot learning from internet-scale human videos.

Abstract

Learning robotic manipulation from human videos is a promising solution to the data bottleneck in robotics, but the distribution shift between humans and robots remains a critical challenge. Existing approaches often produce entangled representations, where task-relevant information is coupled with human-specific kinematics, limiting their adaptability. We propose a generative framework for cross-embodiment video editing that directly addresses this by learning explicitly disentangled task and embodiment representations. Our method factorizes a demonstration video into two orthogonal latent spaces by enforcing a dual contrastive objective: it minimizes mutual information between the spaces to ensure independence while maximizing intra-space consistency to create stable representations. A parameter-efficient adapter injects these latent codes into a frozen video diffusion model, enabling the synthesis of a coherent robot execution video from a single human demonstration, without requiring paired cross-embodiment data. Experiments show our approach generates temporally consistent and morphologically accurate robot demonstrations, offering a scalable solution to leverage internet-scale human video for robot learning.