Text-to-CAD Generation
Overview
Text-to-CAD generation studies how to translate design descriptions into structured models that can be edited and used in engineering workflows. Common approaches have language models generate CAD scripts or feature sequences, often supported by structured representations, multimodal inputs, retrieval, and hierarchical planning. Tool-using agents add execution, geometry checks, and iterative feedback, with some systems also applying reinforcement learning or physical simulation. Models can often produce basic shapes, but reliable complex topology, assemblies, engineering constraints, and generalization remain difficult; executable code alone does not guarantee a correct or usable design.
Latest papers
48 papers2 papers in the last four weeks, down 67% on the four weeks before. 0.0% of all new papers.