Target-Guided Bayesian Flow Networks for Quantitatively Constrained CAD Generation

2025-10-29

AI Summary

Existing methods for multimodal parametric CAD sequence generation struggle to model long-range geometric constraints and suffer from high parameter sensitivity, resulting in low fidelity. Method: This paper proposes the first quantitative-constraint-aware generative framework. Its core innovations are: (1) joint modeling of multimodality within a unified, continuous, and differentiable parameter space; (2) a target-guided Bayesian flow mechanism that encodes geometric constraints—e.g., dimensional accuracy and topological consistency—as differentiable priors for precise attribute control; and (3) integration of diffusion principles with Bayesian update kernels to construct an end-to-end trainable flow network. Results: Evaluated on a custom CAD dataset, our method achieves state-of-the-art performance in both single- and multi-condition generation tasks. Generated CAD sequences strictly satisfy engineering constraints and exhibit significantly improved geometric fidelity.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition
• Machine Learning: Multimodal Learning

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Deep generative models, such as diffusion models, have shown promising progress in image generation and audio generation via simplified continuity assumptions. However, the development of generative modeling techniques for generating multi-modal data, such as parametric CAD sequences, still lags behind due to the challenges in addressing long-range constraints and parameter sensitivity. In this work, we propose a novel framework for quantitatively constrained CAD generation, termed Target-Guided Bayesian Flow Network (TGBFN). For the first time, TGBFN handles the multi-modality of CAD sequences (i.e., discrete commands and continuous parameters) in a unified continuous and differentiable parameter space rather than in the discrete data space. In addition, TGBFN penetrates the parameter update kernel and introduces a guided Bayesian flow to control the CAD properties. To evaluate TGBFN, we construct a new dataset for quantitatively constrained CAD generation. Extensive comparisons across single-condition and multi-condition constrained generation tasks demonstrate that TGBFN achieves state-of-the-art performance in generating high-fidelity, condition-aware CAD sequences. The code is available at https://github.com/scu-zwh/TGBFN.

Problem

• Generating parametric CAD sequences with quantitative constraints
• Handling multi-modal CAD data in unified parameter space
• Addressing long-range constraints and parameter sensitivity issues

Innovation

• Unified continuous space for CAD multi-modality
• Guided Bayesian flow controls CAD properties
• State-of-the-art constrained CAD generation performance