Published Online:July 2026
Product Name:The IUP Journal of Computer Sciences
Product Type:Article
Product Code:IJCS040726
DOI:10.71329/IUPJCS/2026.20.3.43-62
Author Name:Prateek Karna, Rohit Mahato, Pihu Priyansa, Jyotshna Shree, Sagar K G and Yeswanth Kumar S
Availability:YES
Subject/Domain:Engineering
Download Format:PDF
Pages:43-62
The paper synthesizes 48 peer-reviewed studies (2015-2026) selected via a PRISMA-aligned protocol across IEEE Xplore, PubMed Central, Scopus, and Google Scholar, extending a prior corpus of 38 papers with ten new contributions. New additions include a Boeing B- 777 simulator study achieving per-pilot fatigue classification above 95% with multimodal machine learning (ML); a fuzzy-logic kinematic detection scheme requiring no on-body hardware; a deep sparse contractive autoencoder (DCSAEN) achieving 81.5% EEG-based accuracy; a wearable multimodal framework on 27 subjects identifying differential physical and mental fatigue signatures; a rapid EEG-ECG fusion pipeline achieving 88.42% crosssubject accuracy in 39.3 s training time; a PSO-CNN facial feature model achieving 93.9% accuracy; a real-world fatigue management technology (FMT) field trial; and forensic speechacoustic analysis validating zero-hardware fatigue detection. Advanced architectures covered include STG-CLNet (95.6%), FatigueNet GNN-Transformer (90.2%), DRN-RF (93.1%), and the ASFT-Transformer (88.01%). Integration of explainable AI (XAI), generative AI (GAI), and human-centric artificial cognitive system (ACS) frameworks is discussed toward a trusted and real-time aviation safety ecosystem.
As the aviation industry moves toward more automated cockpits and single-pilot operations, the focus for human operators has shifted from physical control to ongoing monitoring.