发表机构
The Australian National University; Mitsubishi Electric Research Laboratories (MERL); Friedrich-Alexander-Universität Erlangen-Nürnberg; Tsinghua University（澳大利亚国立大学; 三菱电机研究实验室（MERL）; 弗里德里希-亚历山大大学埃尔朗根-纽伦堡; 清华大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出AssemblyWorld交互式3D装配环境及AssemblyWorldBench基准，评估八个通用智能体，发现最强系统零件准确率80.9%但完整装配成功率仅59.4%，开源系统明显落后于闭源系统。
AI 中文摘要
3D装配任务要求将零件及其关系的理解转化为精确的空间排列。预训练的通用智能体能否通过视觉交互来装配物体，而无需额外的装配专用微调？为了探究这一问题，我们引入了AssemblyWorld，一个交互式3D环境，其中智能体检查渲染视图并操作提供的刚性零件，在可用时由图像或装配手册引导。智能体通过2D视图感知零件几何形状，而非直接访问网格顶点或面，其最终装配结果则通过几何方式进行评估。基于该环境，我们构建了AssemblyWorldBench，包含跨越家具、工业装配和断裂重组等80个物体的100个装配任务。对八个智能体系统的评估揭示了它们在能力上的显著差异。最强的系统实现了80.9%的零件准确率，但完整装配成功率仅为59.4%。被评估的开源系统在执行可靠性和装配准确性方面均明显落后于其更强大的闭源同类系统。对视觉参考、交互轨迹和失败的分析显示了智能体如何修订装配方案，同时留下残余定位误差。AssemblyWorld为评估交互式装配智能体的能力以及刻画近似结构恢复与精确重建之间的差距提供了一个共同的环境。
英文摘要
The task of 3D assembly requires translating an understanding of parts and their relationships into precise spatial arrangements. Can pretrained general-purpose agents assemble objects through visual interaction without additional assembly-specific fine-tuning? To investigate this question, we introduce AssemblyWorld, an interactive 3D environment in which agents inspect rendered views and manipulate supplied rigid parts, guided by images or assembly manuals when available. Agents perceive part geometry through 2D views rather than direct access to mesh vertices or faces, while their resulting assemblies are evaluated geometrically. Building on this environment, we construct AssemblyWorldBench, comprising 100 assembly tasks across 80 objects spanning furniture, industrial assembly, and fracture reassembly. Evaluating eight agent systems reveals substantial differences in their capabilities. The strongest system achieves 80.9% part accuracy but 59.4% complete-assembly success. The evaluated open-source systems lag substantially behind their stronger closed-source peers in both execution reliability and assembly accuracy. Analyses of visual references, interaction trajectories, and failures show how agents revise assemblies while leaving residual positioning errors. AssemblyWorld provides a common setting for both assessing the capabilities of interactive assembly agents and characterizing the gap between approximate structure recovery and precise reconstruction.
Comments24 pages, 11 figures. Project page: https://assemblyworld.github.io
