Bridging Design Gaps: A Parametric Data Completion Approach With Graph Guided Diffusion Models

TL;DR

The paper tackles the challenge of imputing missing parametric data in engineering design by integrating diffusion models with graph neural networks guided by assembly graphs, framing the task as an AI design copilot that can propose multiple complete designs. It introduces a five-step end-to-end pipeline that tokenizes multimodal features, fuses graph- and tabular-conditioned information via cross-attention, and uses a diffusion denoising process to generate complete parametric designs, which are then rendered. Through extensive experiments on an augmented bicycle CAD dataset, the approach beats classical deterministic imputers (e.g., MissForest, hotDeck, PPCA) and a leading diffusion-based method (TabCSDI) in RMSE, error rate, and design diversity, while preserving realistic feature distributions as shown by KL-divergence analyses. The results demonstrate that the model can serve as a credible design copilot, offering diverse, high-fidelity design completions that facilitate ideation and informed decision-making, with potential to generalize to other CAD domains and engineering fields.

Abstract

This study introduces a generative imputation model leveraging graph attention networks and tabular diffusion models for completing missing parametric data in engineering designs. This model functions as an AI design co-pilot, providing multiple design options for incomplete designs, which we demonstrate using the bicycle design CAD dataset. Through comparative evaluations, we demonstrate that our model significantly outperforms existing classical methods, such as MissForest, hotDeck, PPCA, and tabular generative method TabCSDI in both the accuracy and diversity of imputation options. Generative modeling also enables a broader exploration of design possibilities, thereby enhancing design decision-making by allowing engineers to explore a variety of design completions. The graph model combines GNNs with the structural information contained in assembly graphs, enabling the model to understand and predict the complex interdependencies between different design parameters. The graph model helps accurately capture and impute complex parametric interdependencies from an assembly graph, which is key for design problems. By learning from an existing dataset of designs, the imputation capability allows the model to act as an intelligent assistant that autocompletes CAD designs based on user-defined partial parametric design, effectively bridging the gap between ideation and realization. The proposed work provides a pathway to not only facilitate informed design decisions but also promote creative exploration in design.

Figures (8)

• Figure 2: Diagram shows TabCSDI's model architecture.
• Figure 3: Overview of our model.
• Figure 4: Structure of bike assembly graph.
• Figure 5: Overall Distribution study for three selected features that were missing---handle bar style, saddle height, and stem angle
• Figure 6: Distribution study of feature stem angle for two randomly chosen bikes. Dark red lines represent the average value for stem angle among all the generated examples and the dark blue lines represent the dataset value for stem length for the testing bikes being studied. The "dist" number indicates the average distance between the dark red lines and the dark blue lines.