Review possible WAM paper: DVG-WM: Disentangled Video Generation Enables Efficient Embodied World Model for Robotic Manipulation

[279object]

Paper

• Title: DVG-WM: Disentangled Video Generation Enables Efficient Embodied World Model for Robotic Manipulation
• Short name: DVG-WM
• arXiv ID: 2606.32028v1
• Paper: https://arxiv.org/pdf/2606.32028
• Authors: Ziyu Shan, Zhenyu Wu, Xiaofeng Wang, Zheng Zhu, Ziwei Wang
• Published: 2026-06-30
• arXiv categories: cs.RO
• Matched keywords: embodied world model, embodied world models

README Entry

- **DVG-WM**: "DVG-WM: Disentangled Video Generation Enables Efficient Embodied World Model for Robotic Manipulation", arXiv 2026. ![](https://img.shields.io/badge/Needs--Taxonomy--Review-64748b)
  [[📄 Paper](https://arxiv.org/pdf/2606.32028)]

Robot Decision

• Decision: pending
• Confidence: medium
• Taxonomy: None
• Taxonomy confidence: None
• README heading: ## World Action Model
• Badges: https://img.shields.io/badge/Needs--Taxonomy--Review-64748b

Reason

Embodied robotic world model, but the proposed model is language/goal-conditioned rather than clearly action-conditioned.

Evidence

• predicts future observations from initial frame and language instruction
• paper explicitly says it focuses on goal-conditioned setting
• uses inverse dynamics/action expert after video prediction

Needs taxonomy review: fallback README heading/badge was used because taxonomy.toml has no README mapping for this classification.

Human Review Checklist

• This paper belongs in the WAM survey
• README section is correct
• Badges are correct
• Short name is correct
• Paper link is correct
• Merge this PR if accepted