RLCAD: Reinforcement Learning Training Gym for Revolution Involved CAD Command Sequence Generation

2025-03-24

AI Summary

Existing CAD command-sequence generation methods support only 2D sketching, extrusion, and Boolean operations, limiting their ability to model complex geometries involving revolution features. This work introduces the first reinforcement learning (RL) training environment for parametric CAD modeling that natively supports revolution operations, enabling end-to-end, closed-loop generation of fully parameterized modeling commands from B-Rep inputs. We innovatively incorporate revolution as a learnable atomic action within the RL action space and propose a geometric-feedback-driven PPO training paradigm that tightly integrates a CAD kernel, B-Rep difference metrics, state encoding, and action modeling. On the B-Rep-to-command-sequence generation task, our method achieves state-of-the-art performance. Moreover, it improves training efficiency by 39× over prior CAD RL environments, demonstrating significant advances in both expressiveness and scalability for learning-based CAD synthesis.

Technology Category

• Machine Learning: Reinforcement Learning
• Natural Language Processing: Generation
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

A CAD command sequence is a typical parametric design paradigm in 3D CAD systems where a model is constructed by overlaying 2D sketches with operations such as extrusion, revolution, and Boolean operations. Although there is growing academic interest in the automatic generation of command sequences, existing methods and datasets only support operations such as 2D sketching, extrusion,and Boolean operations. This limitation makes it challenging to represent more complex geometries. In this paper, we present a reinforcement learning (RL) training environment (gym) built on a CAD geometric engine. Given an input boundary representation (B-Rep) geometry, the policy network in the RL algorithm generates an action. This action, along with previously generated actions, is processed within the gym to produce the corresponding CAD geometry, which is then fed back into the policy network. The rewards, determined by the difference between the generated and target geometries within the gym, are used to update the RL network. Our method supports operations beyond sketches, Boolean, and extrusion, including revolution operations. With this training gym, we achieve state-of-the-art (SOTA) quality in generating command sequences from B-Rep geometries. In addition, our method can significantly improve the efficiency of command sequence generation by a factor of 39X compared with the previous training gym.

Problem

• Extends CAD command sequence generation to include revolution operations
• Improves geometric complexity representation in automated CAD design
• Enhances efficiency and quality of reinforcement learning-based CAD generation

Innovation

• Reinforcement learning gym for CAD command generation
• Supports revolution operations beyond basic sketching
• Achieves 39X efficiency improvement in sequence generation