Artificial Intelligence for first-time-right Additive Manufacturing in the defence sector: novel solutions and current perspective

Additive Manufacturing (AM) has emerged as a critical enabling technology across various industries, particularly in the defense and aerospace sectors. In these fields, "first-time-right" production is essential, as manufacturing often occurs in situ with limited operator expertise, requiring maximum efficiency, minimal waste, and short production times. For defense applications, produced parts must meet stringent quality specifications without relying on iterative trial-and-error processes. In this context, in-situ and in-line monitoring of the AM process, augmented by Artificial Intelligence (AI) methods, presents novel opportunities and significant advancements. Layer-by-layer big data streams acquired during production enable fast and efficient detection of process anomalies and product defects. This study explores AI-driven solutions developed and validated at Politecnico di Milano to facilitate first-time-right AM of complex geometries and mission-critical components. Key focus areas include the integration of machine learning models for real-time monitoring using layerwise image data, high-speed video streams, and thermal maps gathered through in-situ thermography. Real-world case studies are presented to demonstrate the benefits of the proposed methods in automated anomaly detection and their readiness for industrial adoption. Additionally, we discuss emerging AI techniques and the associated opportunities in AM, spanning from design to maintenance. The current perspective on AI adoption in AM for defense applications is reviewed, highlighting barriers and potential strategies to overcome them.