Towards High-Fidelity CAD Generation via LLM-Driven Program Generation and Text-Based B-Rep Primitive Grounding

2026-03-12

AI Summary

This work addresses the paradigmatic divide between parametric modeling and boundary representation (B-Rep) in existing AI-driven CAD systems, which hinders high-fidelity design of complex industrial products. To bridge this gap, we propose FutureCAD, a novel framework that explicitly links large language models (LLMs) with B-Rep geometry. FutureCAD leverages an LLM to generate executable CadQuery scripts and integrates a text-query mechanism with a B-Rep grounding transformer to enable precise mapping from natural language to geometric primitives, thereby unifying parametric operations and B-Rep representations. A key innovation is the introduction of a text-driven geometric primitive selection mechanism that effectively reconciles the two modeling paradigms. Evaluated on a newly curated real-world CAD dataset and trained via supervised fine-tuning and reinforcement learning, FutureCAD achieves state-of-the-art performance in high-fidelity CAD generation, significantly improving modeling accuracy and generalization.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Computational Creativity
• Computer Vision: Large Vision Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

The field of Computer-Aided Design (CAD) generation has made significant progress in recent years. Existing methods typically fall into two separate categorie: parametric CAD modeling and direct boundary representation (B-Rep) synthesis. In modern feature-based CAD systems, parametric modeling and B-Rep are inherently intertwined, as advanced parametric operations (e.g., fillet and chamfer) require explicit selection of B-Rep geometric primitives, and the B-Rep itself is derived from parametric operations. Consequently, this paradigm gap remains a critical factor limiting AI-driven CAD modeling for complex industrial product design. This paper present FutureCAD, a novel text-to-CAD framework that leverages large language models (LLMs) and a B-Rep grounding transformer (BRepGround) for high-fidelity CAD generation. Our method generates executable CadQuery scripts, and introduces a text-based query mechanism that enables the LLM to specify geometric selections via natural language, which BRepGround then grounds to the target primitives. To train our framework, we construct a new dataset comprising real-world CAD models. For the LLM, we apply supervised fine-tuning (SFT) to establish fundamental CAD generation capabilities, followed by reinforcement learning (RL) to improve generalization. Experiments show that FutureCAD achieves state-of-the-art CAD generation performance.

Problem

• CAD generation
• parametric modeling
• boundary representation
• geometric primitives
• AI-driven design

Innovation

• LLM-driven CAD generation
• B-Rep grounding
• text-to-CAD
• parametric modeling
• geometric primitive selection