Explore
Automated Optical Inspection of Deformable Linear Objects in Wiring Harnesses Applying Deep Learning and Synthetic Data
Huong Giang Nguyen
2026
0 Ungluers have
Faved this Work
Login to Fave
Automated optical inspection (AOI) systems are an important building block in intelligent manufacturing systems. They enable the acquisition of optical data for process monitoring and documentation, while also facilitating process optimization through data analysis. The wiring harness industry, in particular, poses challenges due to a high-variant manufacturing philosophy, manual assembly processes, and quality requirements. In this context, AOI systems offer significant potential by enabling automated quality control, fast failure detection, and trouble-shooting of manual processes. This thesis introduces the FastAOI methodology, which provides a systematic approach for developing smart AOI systems using data driven techniques and synthetic data. FastAOI comprises the process steps AOI planning, 3D product model creation, scene creation, and data processing. The approach leverages deep learning algorithms trained with synthetic data to detect wiring harness components, including both rigid and deformable linear objects. FastAOI is implemented and vali-dated in two AOI systems, one for image and one for point cloud seg-mentation, to demonstrate its applicability across multiple data modalities. The research results confirm the viability of deep learning for high-performance wiring harness segmentation. Furthermore, effective strategies for synthetic data integration into deep learning training and the positive impact of synthetic data on the model performance are outlined.
This book is included in DOAB.
Rights Information
Are you the author or publisher of this work? If so, you can claim it as yours by registering as an Unglue.it rights holder.Downloads
This work has been downloaded 0 times via unglue.it ebook links.
- 0 - pdf (CC BY-NC) at OAPEN Library.
Keywords
- automated optical inspection
- deep learning
- Deformable Linear Objects
- machine vision
- synthetic data
- wiring harness
Links
DOI: 10.25593/978-3-96147-922-1Editions