🤖 AI Summary
This work addresses the frequent lack of physical plausibility in CAD designs generated by large language models, which often fail to meet engineering reliability standards. To overcome this limitation, the authors propose a hybrid agent-physical architecture that, for the first time, deeply integrates explicit physics-based validation into the decision-making loop of generative CAD agents. The approach employs an iterative pipeline of planning, generation, evaluation, and revision, combining a large language model–driven multi-agent system with knowledge engineering tools to enable structural simulation and verification. Experimental results demonstrate that the proposed method yields designs with 4.2× higher structural complexity and a 3.5% improvement in compilation success rate. The authors have open-sourced the code, prompts, and dataset to support reproducible research.
📝 Abstract
Large Language Models (LLMs) can generate Computer-Aided Design (CAD), yet lack physical comprehension required for reliable engineering design. Instead of attempting to implicitly learn physical laws from data, we propose a Hybrid Agentic-Physical Architecture that embeds validated knowledge-based engineering tools directly into the decision making loop of autonomous AI agents. In this framework, engineering design is formulated as a closed-loop, sequential decision making process guided by explicit physical verification. Based on a load case, dedicated agents iteratively plan, generate, evaluate, and revise engineering designs using knowledge-based tools as a feedback signal. We introduce a benchmark dataset and metrics for assessing functional validity in generative CAD. Our system generates more complex and physically verified designs, with a 4.2 increase in structural complexity and improving compile rate by 3.5% compared to similar agentic methods. The codebase, prompts and dataset will be made publicly available to support reproducibility and future research.