The paper introduces RS-CNN (Relation-Shape Convolutional Neural Network), a novel approach for point cloud analysis that extends regular grid CNN to handle irregular configurations. The key innovation is the relation-shape convolutional operator, which learns from the geometric topology constraints among points, encoding meaningful shape information. This operator allows RS-CNN to capture explicit spatial layouts of points, leading to shape-aware and robust representations. The hierarchical architecture of RS-CNN enables contextual shape-aware learning, making it highly effective for various tasks such as shape classification, part segmentation, and normal estimation. Extensive experiments on challenging benchmarks demonstrate that RS-CNN outperforms state-of-the-art methods, achieving state-of-the-art results in multiple tasks. The paper also includes a detailed analysis of the design choices, ablation studies, and robustness to point permutation and rigid transformations, providing a comprehensive evaluation of the proposed method.The paper introduces RS-CNN (Relation-Shape Convolutional Neural Network), a novel approach for point cloud analysis that extends regular grid CNN to handle irregular configurations. The key innovation is the relation-shape convolutional operator, which learns from the geometric topology constraints among points, encoding meaningful shape information. This operator allows RS-CNN to capture explicit spatial layouts of points, leading to shape-aware and robust representations. The hierarchical architecture of RS-CNN enables contextual shape-aware learning, making it highly effective for various tasks such as shape classification, part segmentation, and normal estimation. Extensive experiments on challenging benchmarks demonstrate that RS-CNN outperforms state-of-the-art methods, achieving state-of-the-art results in multiple tasks. The paper also includes a detailed analysis of the design choices, ablation studies, and robustness to point permutation and rigid transformations, providing a comprehensive evaluation of the proposed method.