EUVIMEDOpdag EUVIMED
Crossref

Automated Defect Recognition in Solid Rocket Motor Manufacturing: Reconciling AI Performance Metrics with ECSS POD Requirements for Digital Radiographic Inspection of Solid Rocket Motors

Domenico Telesca, Alessio Franchina, Liberata Furente

e-Journal of Nondestructive Testing · 2026-07 · 31 (7)

DOI: 10.58286/33590 ↗

At a glance

Journal article in e-Journal of Nondestructive Testing, 2026-07. Written by Domenico Telesca and 2 co-authors.

AbstractBibliografiske oplysningerCiteRelated studies

Abstract (original language)

The aerospace industry faces a critical challenge in qualifying artificial intelligence systems for safety-critical inspection: reconciling machine learning performance metrics with rigorous Probability of Detection (POD) requirements mandated by ECSS standards. This paper presents the development and qualification of an AI-powered defect detection system for digital radiographic inspection of solid rocket motors at AVIO S.p.A. Our work addresses the fundamental tension between AI validation practices (precision, recall, F1-score) and aerospace POD requirements (90% detection probability at 95% confidence through hit/miss analysis). Key challenges include complex component geometries, multiple material interfaces, limited training datasets due to specimen fabrication costs, and integration with existing qualified inspection workflows. We demonstrate that AI recall metrics can support, but not replace, formal POD validation on physical specimens. Case study in AVIO is presented with the aim of 80% reduction in inspection time while maintaining human-in-the-loop final authority. The study provides practical guidance for aerospace organizations pursuing AI-assisted NDT qualification under European standards.

Cite

APA

Telesca, D., Franchina, A., & Furente, L. (2026). Automated Defect Recognition in Solid Rocket Motor Manufacturing: Reconciling AI Performance Metrics with ECSS POD Requirements for Digital Radiographic Inspection of Solid Rocket Motors. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/33590

Vancouver

Telesca D, Franchina A, Furente L. Automated Defect Recognition in Solid Rocket Motor Manufacturing: Reconciling AI Performance Metrics with ECSS POD Requirements for Digital Radiographic Inspection of Solid Rocket Motors. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/33590

BibTeX
@article{Telesca2026,
  author = {Telesca, Domenico and Franchina, Alessio and Furente, Liberata},
  title = {Automated Defect Recognition in Solid Rocket Motor Manufacturing: Reconciling AI Performance Metrics with ECSS POD Requirements for Digital Radiographic Inspection of Solid Rocket Motors},
  journal = {e-Journal of Nondestructive Testing},
  year = {2026},
  volume = {31},
  number = {7},
  doi = {10.58286/33590}
}
Start for free →