# Hierarchical Error Assessment of CAD Models for Aircraft Manufacturing-and-Measurement

arXiv: https://arxiv.org/abs/2506.10594 · HTML: https://mlsanity.com/paper/2506.10594/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 33 references that are in the map.
- First posted on arXiv 12 Jun 2025.

## Builds on (most cited first, 7 of 7)

- [PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space](https://mlsanity.com/paper/1706.02413/) (2017)
- [ABC: A Big CAD Model Dataset For Geometric Deep Learning](https://mlsanity.com/paper/1812.06216/) (2018)
- [DeepICP: An End-to-End Deep Neural Network for 3D Point Cloud Registration](https://mlsanity.com/paper/1905.04153/) (2019)
- [PCRNet: Point Cloud Registration Network using PointNet Encoding](https://mlsanity.com/paper/1908.07906/) (2019)
- [HPNet: Deep Primitive Segmentation Using Hybrid Representations](https://mlsanity.com/paper/2105.10620/) (2021)
- [ImLoveNet: Misaligned Image-supported Registration Network for Low-overlap Point Cloud Pairs](https://mlsanity.com/paper/2207.00826/) (2022)
- [Searching for Representative Modes on Hypergraphs for Robust Geometric Model Fitting](https://mlsanity.com/paper/1802.01129/) (2018)

## Abstract (arXiv)

The most essential feature of aviation equipment is high quality, including high performance, high stability and high reliability. In this paper, we propose a novel hierarchical error assessment framework for aircraft CAD models within a manufacturing-and-measurement platform, termed HEA-MM. HEA-MM employs structured light scanners to obtain comprehensive 3D measurements of manufactured workpieces. The measured point cloud is registered with the reference CAD model, followed by an error analysis conducted at three hierarchical levels: global, part, and feature. At the global level, the error analysis evaluates the overall deviation of the scanned point cloud from the reference CAD model. At the part level, error analysis is performed on these patches underlying the point clouds. We propose a novel optimization-based primitive refinement method to obtain a set of meaningful patches of point clouds. Two basic operations, splitting and merging, are introduced to refine the coarse primitives. At the feature level, error analysis is performed on circular holes, which are commonly found in CAD models. To facilitate it, a two-stage algorithm is introduced for the detection of circular holes. First, edge points are identified using a tensor-voting algorithm. Then, multiple circles are fitted through a hypothesize-and-clusterize framework, ensuring accurate detection and analysis of the circular features. Experimental results on various aircraft CAD models demonstrate the effectiveness of our proposed method.