PRECISION MANUFACTURING IN AEROSPACE ENGINEERING USING ADVANCED ROBOTIC SYSTEMS

Abstract

Automated inspection and robotic process control of the manufacturing process are becoming more integral to the precision aerospace manufacturing process to ensure product quality and assembly consistency. In this study, a complementary framework based on two modules, a robotic fastener-process characterization module and an artificial-intelligence-based aircraft-defect-detection module, were developed. The first module involved analyzing 479 trials of the KUKA KR16 for aeronautical fastener-assemblies based on temporal kinematic and dynamic signals, engineering indicators, deviation measures, and supervised learning for assessment of the assembly outcome. The second module used a leakage-controlled YOLO11n workflow for dent, fastener damage and rupture detection, validation only model selection, locked testing, robustness analysis, bootstrapping uncertainty estimation and object-scale diagnostics and checkpoint-reproducibility controls. There was still the 3-class model with a Macro-F1 of 0.4127, but none of the engineering KPI families were of significance following FDR correction. The chosen YOLO11n-416 detector performed best with test precision of 0.8027, recall of 0.5612 and mAP50–95 of 0.2993, substantially compromised by the performance on the small defects. These results suggest the use of complementary evidence of process stability and quality verification, and not just a single performance measure, to assess the precision of the aerospace manufacturing industry. The framework lays the groundwork for a repeatable approach to the integration of robotic execution analysis with uncertainty-aware visual quality assessment without baseless claims of actual direct robotic-vision fusion or proven dimensional accuracy gains in production.