PointWAM: 3D World Action Modeling for Dexterous Robotic Manipulation

• 类型：arxiv
• 标识：2610.02840
• 链接：https://arxiv.org/abs/2610.02840
• 主分类：engineering
• 形态：method
• TLDR：World action models jointly learn to forecast world dynamics and predict robot actions, such that the learned internal world dynamics guide accurate actions. Existing approaches typically represent the world as RGB frames or latent counterparts while predicting actions as end-effector poses or joint angles, but they often struggle to capture the 3D spatial structure and contact geometry central to dexterous manipulation. We introduce Point World Action Model (PointWAM), a 3D world action model that decomposes the world into a scene (i.e., environment) and hands (i.e., actor), and jointly forec
• 待LLM分类：是
• 标题中文：PointWAM：面向灵巧机器人操作的 3D World Action Modeling
• TLDR中文：World action models 联合学习预测世界动态与机器人动作，使所学内部世界动态指导精确动作。现有方法通常将世界表示为 RGB 帧或对应隐空间，并将动作预测为末端执行器位姿或关节角度，但难以捕捉对灵巧操作至关重要的 3D 空间结构与接触几何。本文提出 Point World Action Model (PointWAM)，这是一种将世界分解为场景（即环境）与手部（即执行者）的 3D world action model，并联合预测
• 来源文件：
• /inbox/tom/_candidates/2026-10-06-rag-retrieval-reranking-candidates.json