[论文] Towards Physically Grounded Articulated 3D Assets: A Multi-Agent Pipel...

论文概要

研究领域: CV 作者: Xudong Cai, Ruijie Zhu, Haoliang Shang, Ruining Li, Marc Pollefeys, Shangzhe Wu, Andrea Vedaldi, Chuanxia Zheng 发布时间: 2026-10-09 arXiv: 2610.13947

中文摘要

本文研究物理接地（physically grounded）的关节式 3D 资产生成——目标从「复刻物体的外观」转向「捕捉物体的运动与功能」。受程序化生成与智能体生成文献启发，提出 Artisan 系统，在 Articraft 方法基础上引入带物理仿真反馈的多智能体流水线，支持几何、外观、物理可信关节运动之外的综合属性，如无碰撞运动、重力稳定性、准确的质量与材质属性。构建统一基准，覆盖空间保真度、物理可信度、下游任务性能和整体质量四个维度，包含 100 个参考资产、50 个稳定性提示和 50 个带参考质量的 ABO 物体。对 10 余个代表性模型的广泛评估表明：Artisan 在空间保真度上具有竞争力，同时碰撞率最低、稳定率最高。系统与基准共同为推进 3D 生成走向物理可信、功能可用的资产提供了方向。

原文摘要

In this paper, we study physically grounded articulated 3D asset generation---moving the goal from reproducing how objects look to capturing how they move and function. Inspired by the literature on procedural and agentic generation, we present Artisan, a system that extends the latest method Articraft by introducing a multi-agent pipeline with physical simulation feedback, enabling comprehensive attributes beyond geometry, appearance, and physically plausible articulation, such as collision-free motion, stability under gravity, and accurate mass and material properties. Furthermore, we construct a unified benchmark spanning four key aspects: spatial fidelity, physical plausibility, downstream task performance, and overall quality. The benchmark encompasses 100 reference assets from ArtVIP...

#论文 #arXiv #CV #小凯