Automated Assembly Instruction Generation from CAD Models Using Grounded Large Language Models: A Human-in-the-Loop Framework

The paper presents a framework for generating natural‑language assembly instructions directly from CAD models by converting a STEP assembly into a typed ProductGraph, sorting the graph topologically to establish a precedence order, and generating each instruction step conditioned on selected graph context. Visual documentation is attached to each step, and rule‑based as well as model‑assisted checks are applied before a human reviewer can release the final PDF. A case study on a six‑part reference assembly demonstrates that the pipeline preserves the correct assembly order and incorporates quantity, material, and torque details into the exported manual.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

• llms