BrickWright 用智能体生成物理可行的乐高模型
AI 总结
Ava Pun 等十位作者在 arXiv 发表的论文提出 BrickWright,利用预训练编程智能体根据文本或图像提示生成物理可行的乐高积木模型。其 BrickLayer Python 工具包包含积木装配 API 和物理可行性验证器。作者报告称,BrickWright 引导的智能体在 512 个提示样本上达到 100% 有效率,而基线结果从 GPT-6 Luna 的 9% 到 GPT-6 Astra 的 69% 不等。
深度见解
论文报告称,装配与验证工具能帮助编程智能体避免出现悬空、碰撞或断开的零件,同时生成彼此连接并符合输入提示的结构。
Computer Science > Computer Vision and Pattern Recognition
arXiv:2610.31651 (cs)
[Submitted on 22 Oct 2026]
Title:BrickWright: Agentic Creation of Physically Feasible Brick Builds
Authors:Ava Pun, Kevin Dai, Sally Chen, Maxwell Jones, Ruixuan Liu, Kangle Deng, Changliu Liu, Maneesh Agrawala, Deva Ramanan, Jun-Yan Zhu
View a PDF of the paper titled BrickWright: Agentic Creation of Physically Feasible Brick Builds, by Ava Pun and 9 other authors