面向全六面体网格的CAD分解拓扑感知强化学习
Topology-Aware Reinforcement Learning for CAD Decomposition toward All-Hexahedral Meshing
Jiaming Peng · Zhihao Zheng · Rui Wang · Qinglin Wang · Jing Xiao · Jie Liu · et al.
中文摘要
复杂CAD模型的自动六面体网格生成长期被视为网格生成领域的“圣杯”。核心挑战来自特征错位:正交或交互特征共存,使直接六面体网格生成不可行,需要分解为更简单、可网格化的块。现有分解方法(包括框架场流水线和基于学习的技术)虽减少了人工干预,但仍对几何与拓扑复杂度敏感,在困难特征配置下可能无法产生有效分解。其动作空间通常局限于基于表面的切分,常导致无效或不可网格化的分解。为此提出DecompRL,一个面向六面体网格生成的CAD分解深度强化学习框架。设计一组拓扑感知的内部切分动作,利用全局拓扑结构与局部几何属性有效分离错位特征;引入几何奖励与可网格化反馈,引导学习走向有效分解和高质量网格。在ABC、Fusion360、NIST和DeepCAD上的实验表明,DecompRL在四个基准中的三个上取得所评估学习方法中最高的成功率,同时保持有竞争力的网格质量。实测LMFF与DecompRL的运行时间比最高达两个数量级;该比较为完整LMFF流程与仅DecompRL分解(不含后续网格生成)。最大成功率提升出现在Fusion360,DecompRL达到59.6%的成功率。
关键要点
- 01问题:现有CAD分解动作空间局限于表面切分,在特征错位时易产生无效或不可网格化分解。
- 02方法:DecompRL用拓扑感知内部切分动作,结合全局拓扑与局部几何分离错位特征。
- 03训练:引入几何奖励与可网格化反馈,引导强化学习产出有效分解与高质量网格。
- 04结果:在ABC、Fusion360、NIST、DeepCAD中三个基准上成功率最高,Fusion360达59.6%。
- 05效率:LMFF与DecompRL运行时间比最高达两个数量级,仅比较分解阶段。
解读
尚无解读。
原始英文摘要
arXiv:2611.12930v1 Announce Type: cross Abstract: Automatic hexahedral meshing for complex CAD models has long been regarded as the "holy grail" of mesh generation. The fundamental challenge arises from feature misalignment, where orthogonal or interacting features coexist, making direct hexahedral meshing infeasible and requiring decomposition into simpler, meshable blocks. While existing decomposition methods, including frame-field pipelines and learning-based techniques, have reduced manual intervention, they can remain sensitive to geometric and topological complexity and may fail to produce valid decompositions for challenging feature configurations. Specifically, their action spaces are typically restricted to surface-based splitting, which often yields invalid or unmeshable decompositions. To address these limitations, we propose DecompRL, a deep reinforcement learning framework for CAD decomposition toward hexahedral meshing. We design a set of topology-aware internal-splitting actions that leverage both global topological structure and local geometric property to effectively separate misaligned features. Furthermore, we introduce geometric rewards together with meshability feedback to guide the learning process toward valid decompositions and high-quality meshes. Extensive experiments on ABC, Fusion360, NIST, and DeepCAD show that DecompRL achieves the highest success rate among the evaluated learning-based methods on three of the four benchmarks, while maintaining competitive mesh quality. The measured LMFF-to-DecompRL runtime ratio reaches up to two orders of magnitude; this compares the full LMFF procedure with DecompRL decomposition alone, excluding subsequent mesh generation. The largest SR improvement is observed on Fusion360, where DecompRL reaches a success rate of 59.6%.