ShapeSpace (United Kingdom)

• Introduces 3D-CDC: a novel approach for detecting and classifying design changes in 3D CAD models. • Presents 3D-GCD, a new dataset of geometrically altered CAD components with multi- format visual representations. • Evaluates four deep learning models for classifying engineering relevance of design changes. • Demonstrates that CRNNs achieve strong performance with fewer parameters, supporting efficient AI deployment. • Advances human-AI collaboration in CAD by enabling interpretable and workflow-integrated design change analysis. This paper addresses the challenge of supporting dynamic mechanical design processes in distributed, multi-user environments. Although mechanical CAD is a mature technology, assessing the functional and manufacturing impact of design changes remains a labour-intensive manual task. To address this issue, the authors introduce the concept of “3D CAD Change Detection and Classification” (3D-CDC) and present a novel dataset, “3D-GCD” (3D Geometric Change Dataset), consisting of pairs of 3D CAD components with a variety of technically plausible changes. The dataset represents 3D components in multiple 2D image formats (e.g., grayscale and comparative heatmaps) with renderings generated from single or multiple viewpoints. This dataset is used to train four deep learning architectures—Siamese Networks with convolutional encoders, transformer encoders, 3D Convolutional Networks (3D Conv), and Convolutional Recurrent Neural Networks (CRNN)—to classify the engineering significance of changes to a component’s design. The results indicate that while three architectures (Siamese Networks, transformer encoders, and CRNN) exhibit similar accuracy, the CRNN demonstrates greater consistency and efficiency, requiring fewer trainable parameters for 3D-CDC tasks.

Abstract Many companies offer a range of related products that are constructed using similar components and processes. This enables them to meet customer expectations of product variety while minimizing the overheads (e.g., development and manufacturing costs). To support the management of product variety several indices have been proposed in the literature that measure the degree to which component use is standardized across products within the same product family. However, the derivation of some of these statistics can be laborious to calculate due to the effort required to assemble the necessary information. In this paper, we develop an index more suited to the automated data-mining of a company’s product portfolio, which is derived from the Kullback–Leibler divergence. The new measure provides an easily computed probabilistic measure that can be used to characterize the degree of component reuse within a single product, across a family of products, and at the individual component family level. To illustrate their applications, the indices and several existing measures are calculated for two contrasting product types; using the non-differentiating components of two flat-pack furniture ranges and the components of a range of bicycles.

Blortasia is an abstract art world in the sky where you fly freely through a surreal maze of evolving sculptures.