Title: DREAM: a design assistant for assessing additive manufacturability

Authors: Romain Pinquié; Oscar Fossey; Frédéric Segonds

Addresses: G-SCOP UMR CNRS 7296, Université Grenoble Alpes, CNRS, Grenoble INP, G-SCOP, Grenoble, France ' Laboratoire Conception de Produits et Innovation, EA 3927, Arts et Métiers ParisTech, Paris 75013, France ' Laboratoire Conception de Produits et Innovation, EA 3927, Arts et Métiers ParisTech, Paris 75013, France

Abstract: This paper presents a computational design assistant that supports novice designers willing to evaluate the additive manufacturability of their ideas. The design rules extractor for additive manufacturability (DREAM) uses natural language processing techniques to analyse a textual description contained in an ideation card so as to recommend applicable design rules. The design rules for additive manufacturing are structured in an ontology, and the selection of applicable ones depends on their semantic proximity to sentences of the idea card. The empirical evaluation of DREAM led to the identification of six applicable design rules among which two were false positives, which are due to short sentences containing non-discriminant terms. Future works will concentrate on the evaluation with a larger industrial dataset for continuous improvement. Finally, we will analyse the sketches in the idea card and suggest design improvements based on the opportunities offered by additive manufacturing.

Keywords: design rules; natural language processing; NLP; ontology; knowledge management; design for additive manufacturing; DfAM; design rules extractor for additive manufacturability; DREAM.

DOI: 10.1504/IJPLM.2022.127113

International Journal of Product Lifecycle Management, 2022 Vol.14 No.4, pp.328 - 349

Received: 16 Jan 2022
Received in revised form: 05 Apr 2022
Accepted: 10 Jul 2022
Published online: 22 Nov 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article