英文标题：Bridging Language and Physics: Automated Design of Continuum Robots with Large Language Models
作者：Jingyi Chen, Mohan Zhang, Laura Yao, Yingtai Ni, Jianmin Ji, Jie Peng, Song Wang, Tianlong Chen
arXiv ID：2609.08220 | 分类：cs.RO | 发表：2026-09-08
许可：CC-BY
摘要 大语言模型（LLMs）近来已成为一种有前景的工具，能够根据高层级规格自动完成机器人设计，然而对于在复杂物理交互下运行的机器人，它们仍然效果不佳。这一局限源于基于语言的推理与具身化所带来的物理后果之间的鸿沟，往往导致设计在物理有效性上偏低。{{NL}}在本工作中，我们提出了一个多层框架 AID-SR，它通过将仿真器观测到的物理状态转化为面向 LLM 设计者的结构化反馈，从而建立起闭环。结合语义批评、人类反馈与迭代精化，该框架促进了物理上可行且功能上有意义的机器人设计的生成。{{NL}}我们在一个涵盖到达、抓取、运动和操作的 14 项任务基准上，对腱驱动连续体机器人评估了我们的方法。{{NL
Large language models (LLMs) have recently emerged as a promising tool for automating robot design from high-level specifications, yet they remain ineffective for robots operating under complex physical interactions. This limitation stems from the gap between language-based reasoning and the physical consequences of embodiment, often resulting in designs with low physical validity. In this work, we propose a multi-layered framework, AID-SR, that establishes a closed loop by translating simulator-observed physical states into structured feedback for the LLM designer. Combined with semantic critique, human feedback, and iterative refinement, the framework promotes the generation of physically feasible and functionally meaningful robot designs. We evaluate our approach on tendon-driven continuum robots across a benchmark of 14 tasks spanning reaching, grasping, locomotion, and manipulation. The proposed framework achieves 96.2% rate for passing the simulation feasibility check and by applying a common reinforcement learning training, 26.7% robots can successfully fulfill the corresponding task. We then fabricate three designed robots of AID-SR that successfully complete the task in real-world. These extensive experiments across simulation and real-world environments demonstrate and break the wall of utilizing the LLMs for automated design of continuum robots. The source code and experimental resources are publicly available at https://github.com/UNITES-Lab/AID-SR.
查看完整双语翻译 →
正在跳转到翻译阅读页… 如果没有自动跳转，请点击这里。