# DreamX-Phi 1.0：面向机器人操作的动作条件视频世界模型

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-08-13 08:00
- AIHOT 分数：50
- AIHOT 链接：https://aihot.virxact.com/items/cmssf8hxr05tfrod01lrx1icg
- 原文链接：https://arxiv.org/abs/2608.13489

## AI 摘要

DreamX-Phi 1.0 是一个面向机器人操作的动作条件视频世界模型，输入观测帧、语言指令及动作序列即可预测未来观测。该模型通过 PRoPE 式几何编码注入每臂 SE(3) 变换，并引入深度分支与 SAM3 掩码保持物体一致性，同时用分布匹配蒸馏实现高效部署。

## 正文

We present DreamX-Phi 1.0, an action-conditioned video world model for robotic manipulation that, given an observed frame, a language instruction, and a prescribed action sequence comprising end-effector poses and gripper states, predicts the resulting future observations. Yet realism alone does not guarantee faithfulness: a convincing rollout can still move the wrong arm or lose the manipulated object. To ensure the prediction respects each arm's commanded path, we inject per-arm SE(3) transformations into attention via PRoPE-style geometric encoding, preserving arm identity and rigid-motion structure. Action control alone does not fully constrain scene geometry or the evolution of small manipulated objects. We therefore add a lightweight depth branch for scene-level geometry and use SAM3 masks with a frozen V-JEPA teacher to maintain object consistency throughout grasping. We further distill the multi-step generator into a few-step student via distribution-matching distillation for efficient deployment. At the time of writing, achieves first place on Track1 and second place on Track2 of the WorldArena~2.0 Challenge. Our model and code will be publicly available.