The transition toward Industry 5.0 places human-centricity at the core of industrial systems, emphasizing the need to guarantee operator well-being while maintaining productivity and resilience. In visually demanding industrial activities, such as inspection, human–robot collaboration, teleoperation, and augmented reality-assisted assembly, operators are exposed to increasing cognitive demands. Consequently, understanding how mental workload, stress, and related human states are conceptualized, assessed, and integrated into industrial environments has become a critical research challenge. This thesis presents a systematic literature review of empirical studies investigating cognitive and affective state assessment in visually intensive industrial contexts. Adopting a task-centered analytical perspective, the review examines how human states are defined, which indicators are used for their assessment, and to what extent current approaches move beyond descriptive analysis toward automated detection and adaptive integration. Studies were classified according to their methodological maturity, distinguishing descriptive assessment, offline data-driven detection, real-time monitoring, and closed-loop adaptive systems. The findings reveal significant conceptual heterogeneity in construct definitions and substantial variation in measurement strategies. While physiological sensing and machine learning-based detection models are increasingly prevalent, most studies remain limited to offline validation or informational monitoring. Only a small subset implements real-time adaptive mechanisms embedded within industrial architectures. This highlights a persistent gap between state detection and meaningful human-centric system integration. By structuring the field around visually demanding tasks and proposing a maturity-oriented framework, this thesis clarifies the current state of research and identifies structural barriers that hinder the full realization of Industry 5.0 principles in adaptive industrial systems.
La transizione verso l’Industria 5.0 pone la centralità della persona al centro dei sistemi industriali, evidenziando la necessità di garantire il benessere dell’operatore mantenendo al contempo produttività e resilienza. Nelle attività industriali ad elevata richiesta visiva, quali l’ispezione, la collaborazione uomo–robot, la teleoperazione e l’assemblaggio assistito da realtà aumentata, gli operatori sono esposti a crescenti carichi cognitivi. Di conseguenza, comprendere come il mental workload, lo stress e gli stati umani correlati vengano concettualizzati, valutati e integrati negli ambienti industriali rappresenta una sfida di ricerca cruciale. La presente tesi propone una revisione sistematica della letteratura di studi empirici dedicati alla valutazione degli stati cognitivi ed affettivi in contesti industriali ad alta intensità visiva. Adottando una prospettiva analitica centrata sulle attività svolte, la revisione esamina come gli stati umani siano definiti, quali indicatori vengano utilizzati per la loro valutazione e in che misura gli approcci attuali superino l’analisi descrittiva verso il rilevamento automatizzato e l’integrazione adattiva. Gli studi sono stati classificati in base al loro livello di maturità metodologica, distinguendo tra valutazione descrittiva, rilevamento data-driven offline, monitoraggio in tempo reale e sistemi adattivi a ciclo chiuso. I risultati evidenziano una significativa eterogeneità concettuale nelle definizioni dei costrutti e una marcata variabilità nelle strategie di misurazione. Sebbene il monitoraggio fisiologico e i modelli di machine learning siano sempre più diffusi, la maggior parte degli studi rimane limitata a validazioni offline o a sistemi di monitoraggio informativo. Solo una minoranza implementa meccanismi adattivi in tempo reale integrati nelle architetture industriali. Ciò mette in luce un divario persistente tra il rilevamento degli stati dell’operatore e una reale integrazione human-centric dei sistemi produttivi. Attraverso una strutturazione del campo basata sulle attività visivamente impegnative e la proposta di un framework orientato alla maturità metodologica, questa tesi chiarisce lo stato attuale della ricerca e identifica le barriere strutturali che ostacolano la piena realizzazione dei principi dell’Industria 5.0 nei sistemi industriali adattivi.
Assessing mental workload in visual tasks within industry 5.0: a systematic literature review
Normanton Adolpho, Matheus
2024/2025
Abstract
The transition toward Industry 5.0 places human-centricity at the core of industrial systems, emphasizing the need to guarantee operator well-being while maintaining productivity and resilience. In visually demanding industrial activities, such as inspection, human–robot collaboration, teleoperation, and augmented reality-assisted assembly, operators are exposed to increasing cognitive demands. Consequently, understanding how mental workload, stress, and related human states are conceptualized, assessed, and integrated into industrial environments has become a critical research challenge. This thesis presents a systematic literature review of empirical studies investigating cognitive and affective state assessment in visually intensive industrial contexts. Adopting a task-centered analytical perspective, the review examines how human states are defined, which indicators are used for their assessment, and to what extent current approaches move beyond descriptive analysis toward automated detection and adaptive integration. Studies were classified according to their methodological maturity, distinguishing descriptive assessment, offline data-driven detection, real-time monitoring, and closed-loop adaptive systems. The findings reveal significant conceptual heterogeneity in construct definitions and substantial variation in measurement strategies. While physiological sensing and machine learning-based detection models are increasingly prevalent, most studies remain limited to offline validation or informational monitoring. Only a small subset implements real-time adaptive mechanisms embedded within industrial architectures. This highlights a persistent gap between state detection and meaningful human-centric system integration. By structuring the field around visually demanding tasks and proposing a maturity-oriented framework, this thesis clarifies the current state of research and identifies structural barriers that hinder the full realization of Industry 5.0 principles in adaptive industrial systems.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_03_Normanton.pdf
non accessibile
Dimensione
1.74 MB
Formato
Adobe PDF
|
1.74 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/253758