From Intent to Execution: Multimodal Chain-of-Thought Reinforcement Learning for Precise CAD Code Generation

TL;DR

This paper introduces CAD-RL, a multimodal reinforcement learning framework that improves the translation of natural language design intents into precise, executable CAD code, leveraging chain-of-thought reasoning and specialized optimization strategies.

Contribution

It proposes a novel multimodal Chain-of-Thought guided reinforcement learning approach for CAD code generation, along with a new dataset for training and benchmarking.

Findings

CAD-RL outperforms existing models in reasoning and precision.

The framework achieves higher code executability and geometric accuracy.

The dataset ExeCAD supports future research in CAD code generation.

Abstract

Computer-Aided Design (CAD) plays a vital role in engineering and manufacturing, yet current CAD workflows require extensive domain expertise and manual modeling effort. Recent advances in large language models (LLMs) have made it possible to generate code from natural language, opening new opportunities for automating parametric 3D modeling. However, directly translating human design intent into executable CAD code remains highly challenging, due to the need for logical reasoning, syntactic correctness, and numerical precision. In this work, we propose CAD-RL, a multimodal Chain-of-Thought (CoT) guided reinforcement learning post training framework for CAD modeling code generation. Our method combines CoT-based Cold Start with goal-driven reinforcement learning post training using three task-specific rewards: executability reward, geometric accuracy reward, and external evaluation reward. To ensure stable policy learning under sparse and high-variance reward conditions, we introduce three targeted optimization strategies: Trust Region Stretch for improved exploration, Precision Token Loss for enhanced dimensions parameter accuracy, and Overlong Filtering to reduce noisy supervision. To support training and benchmarking, we release ExeCAD, a noval dataset comprising 16,540 real-world CAD examples with paired natural language and structured design language descriptions, executable CADQuery scripts, and rendered 3D models. Experiments demonstrate that CAD-RL achieves significant improvements in reasoning quality, output precision, and code executability over existing VLMs.