An open-source framework for in-process geometric quality assurance in Automated Fiber Placement: A baseline validation and dataset

Automated Fiber Placement (AFP) still lacks reliable and accessible in-process quality assurance methods. As a result, geometric defects such as gaps, overlaps and tow misalignments often remain undetected until post-process inspection, leading to costly scrap or rework. This paper introduces a modular, open-source framework for in-process monitoring and generation of an ”as-built” geometric representation of AFP layups. The system integrates a high-resolution laser line scanner within the AFP head and leverages a direct, high-frequency data link with the robotic arm. An interpretable, Python-based pipeline performs synchronization, feature extraction, 3D coordinate transformation and inter-tow analysis. To establish a baseline validation case, a [0/90] cross-ply planar layup was manufactured and inspected. The framework measured tow-level and interface-level geometric metrics, generated an “as-built” map of the layup and supported localized geometric analysis through an interactive slicing tool. By openly releasing the toolchain and benchmark dataset, this work provides a reproducible, non-proprietary baseline for AFP geometric monitoring research and for future extensions to more complex layups and sensing strategies.