This thesis discloses a method to automatically assess the psycho-physiological status of a pilot over time relying on physiological, behavioral, and environmental measurements. Indeed, as demonstrated by several researches in the literature, human error represents one of the leading causes of aircraft accidents, and stress is the major trigger. Additionally, the increasing complexity of modern combat aircraft contributes at increasing the risk of human error. It follows that such a system represents an effective countermeasure to this issue, by monitoring the pilot status in real-time and leveraging this information to allow an effective fleet management. To reach this objective, a multivariate approach has been implemented, which relies on data collected by a custom hardware setup, ad-hoc designed to acquire physiological, behavioral, and environmental information. In more detail, these signals have been collected for a significant set of aircraft pilots while performing a series of flight maneuvers a priori defined to incrementally become more challenging. These data have been used to design the psycho-physiological index estimation procedure, which relies on unsupervised machine learning techniques. To evaluate the compliance of the produced index with the pilot self-awareness, specific questionnaires have been produced, aimed at investigating their perception concerning the psycho-physiological status during each trial. The results evaluation prove the proposed method to be reliable and robust, paving the way for an in-flight safety system to reduce accidents by being aware of the pilot's psycho-physiological status.
Questa ricerca mira a sviluppare un sistema atto a stimare in modo automatico lo stato psico-fisiologico di un pilot aereonautico sulla base di misure fisiologiche, comportamentali ed ambientali. È infatti dimostrato da molteplici ricerche che gli errori umani sono una delle principali cause di incidenti aerei. Inoltre, la crescente complessità dei sistemi di bordo, che il pilota deve gestire durante il volo, contribuisce ad aumentare il carico di stress ed il conseguente rischio di errore. Ne segue che un sistema capace di monitorare nel tempo lo stato psicofisico del pilota si rivela fondamentale per contribuire a ridurre il rischio di incidenti e migliorare, in generale efficienza e sicurezza in volo. Per raggiungere questo obiettivo, è stato quindi sviluppato un approccio multivariato basato su un setup hardware appositamente progettato per acquisire parametri fisiologici, comportamentali ed ambientali. Più nel dettaglio, è stata definita una procedura basata su tecniche di apprendimento automatico non supervisionato per stimare dai parametri misurtai un indice sintetico dello stato psicofisico. Per validare la bontà dei risultati ottenuti, questi sono stati confrontati con la percezione riportata dagli stess piloti per mezzo di questionari appositamente strutturati e presentati alla fine di ogni test. In conclusione, questa ricerca ha dimostrato con successo il potenziale dell'utilizzo di un approccio multivariato per rilevare automaticamente i cambiamenti nello stato psico-fisiologico dei piloti aeronautici aprendo la strada allo sviluppo di sistemi sempre più sofisticati per migliorare la sicurezza di volo.
PoliMonitor - real-time estimation of a pilot's psychophysical status : a learning-based approach
NOVARO, FEDERICO
2021/2022
Abstract
This thesis discloses a method to automatically assess the psycho-physiological status of a pilot over time relying on physiological, behavioral, and environmental measurements. Indeed, as demonstrated by several researches in the literature, human error represents one of the leading causes of aircraft accidents, and stress is the major trigger. Additionally, the increasing complexity of modern combat aircraft contributes at increasing the risk of human error. It follows that such a system represents an effective countermeasure to this issue, by monitoring the pilot status in real-time and leveraging this information to allow an effective fleet management. To reach this objective, a multivariate approach has been implemented, which relies on data collected by a custom hardware setup, ad-hoc designed to acquire physiological, behavioral, and environmental information. In more detail, these signals have been collected for a significant set of aircraft pilots while performing a series of flight maneuvers a priori defined to incrementally become more challenging. These data have been used to design the psycho-physiological index estimation procedure, which relies on unsupervised machine learning techniques. To evaluate the compliance of the produced index with the pilot self-awareness, specific questionnaires have been produced, aimed at investigating their perception concerning the psycho-physiological status during each trial. The results evaluation prove the proposed method to be reliable and robust, paving the way for an in-flight safety system to reduce accidents by being aware of the pilot's psycho-physiological status.File | Dimensione | Formato | |
---|---|---|---|
Tesi MSc - Federico Novaro.pdf
accessibile in internet per tutti
Descrizione: Tesi
Dimensione
9.66 MB
Formato
Adobe PDF
|
9.66 MB | Adobe PDF | Visualizza/Apri |
Executive Summary -Federico Novaro.pdf
accessibile in internet per tutti
Descrizione: Executive Summary
Dimensione
997.54 kB
Formato
Adobe PDF
|
997.54 kB | 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/10589/209380