Answers are grounded in this paper and cite their sources.
Artificial intelligence (AI) is increasingly being used to support design for manufacturability (DfM) by enabling the generation of manufacturable designs. We refer to this emerging area as generative design for manufacturability (GDfM). Despite growing interest, current literature lacks a dedicated analysis of different approaches for integrating DfM rules in GDfM. This paper presents a systematic review of GDfM based on a PRISMA‐guided search of Scopus, Web of Science, and IEEE Xplore. Of 1547 identified records, 36 studies were selected for detailed analysis. The reviewed studies are compared by manufacturing process, AI method, and data representation. The results show that the literature is dominated by additive manufacturing, with comparatively fewer studies in machining and very limited coverage of casting, molding, forging, and stamping. DfM rules are integrated into designs via data‐driven approaches and rule‐based filters or optimizers within generative design pipelines. Data‐driven approaches ensure manufacturability via manufacturable datasets, enhanced model training, and modified network architectures. Variational autoencoders (VAEs) and generative adversarial networks (GANs) are the most commonly used AI models, whereas other promising models, such as diffusion models and transformers, remain underexplored. In terms of representation, 2D images and voxels dominate despite limited compatibility with downstream engineering tasks. The review identifies key gaps in GDfM in manufacturing process diversity, datasets, benchmarking, and downstream‐compatible representations, and highlights future opportunities in parametric/CAD sequence–based representations, diffusion and transformer models, and broader coverage of traditional manufacturing processes.
No takes yet. Share an insight, caveat, or question.
Weragoda et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: