Metraipolis › archive › dc_f7241bd765d0

Natural-language control of parametric road design based on large language models (2026) · DOI 10.17816/transsyst716698

Source
Metraipolis archive
Original
https://doi.org/10.17816/transsyst716698
License (as archived)
CC-BY-4.0
Archived
2026-10-25T07:14:17.234Z
SHA-256 of the archived text
f77cd38f1e1a13ffe589f3fcb189fe6301a8457442e9691a937361e2a9e53a5d
Stable URL of this page
https://metraipolis.com/documento/dc_f7241bd765d0

How to cite

Natural-language control of parametric road design based on large language models (2026) · DOI 10.17816/transsyst716698 — Metraipolis archive, https://doi.org/10.17816/transsyst716698 (CC-BY-4.0); archived copy: https://metraipolis.com/documento/dc_f7241bd765d0 (Metraipolis archive dc_f7241bd765d0, sha256 f77cd38f1e1a13ffe589f3fcb189fe6301a8457442e9691a937361e2a9e53a5d)
BibTeX
@misc{dcf7241bd765d0,
  title = {Natural-language control of parametric road design based on large language models (2026) · DOI 10.17816/transsyst716698},
  howpublished = {Metraipolis archive},
  url = {https://doi.org/10.17816/transsyst716698},
  note = {License CC-BY-4.0; archived copy https://metraipolis.com/documento/dc_f7241bd765d0; sha256 f77cd38f1e1a13ffe589f3fcb189fe6301a8457442e9691a937361e2a9e53a5d},
  year = {2026}
}

Archived text

Reproduced under CC-BY-4.0. Attribution: Metraipolis archive, https://doi.org/10.17816/transsyst716698. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Articolo recente con licenza aperta: scheda dai metadati Crossref (journal-article).
Titolo: Natural-language control of parametric road design based on large language models
Autori: Andrey A. Kotov
Pubblicato: 2026-09-28 · in: Modern Transportation Systems and Technologies · editore: ECO-Vector LLC (Publications)
DOI: https://doi.org/10.17816/transsyst716698
Licenza dell'articolo: CC-BY-4.0

Abstract (testo degli autori, CC-BY-4.0):
Background. Natural-language control forms a new direction in the development of road computer-aided design systems. However, direct conversion of an engineering task into a program script by a large language model does not guarantee geometric, computational, or regulatory correctness of the result. Aim: To assess the applicability of large language models to the formation of parametric road-object models and to substantiate an architecture for their safe integration into the design process. Materials and methods: In the FreeCAD environment, the creation of a model of a road sign and a multicomponent fragment of a category II road based on text prompts was studied. Script execution, geometry editability, the number of refinement cycles, and the volume of tokens used were evaluated. Results: The fundamental feasibility of natural-language control over a multi-component road model was demonstrated. It was established that a direct generative approach retains significant engineering labor and requires an external validation loop for geometric, computational, and regulatory compliance. A target control architecture and a reconstruction completeness criterion are proposed. Conclusion: The functional separation among the language model, geometric kernel, validation tools, and engineer forms the foundation for a new generation of intelligent road CAD systems.

Fonte: https://api.crossref.org/works/10.17816%2Ftranssyst716698 — metadati Crossref, consultati il 2026-10-25. Interrogazione: «multi-agent systems large language models».