Enhancing flight operations safety : a neurophysiological approach to mental state prediction in commercial single-pilot operations

Abstract

With the potential pilot shortage and economic imperative of cost savings (compelling motivations), Single-Pilot Operations (SPO) is being considered in commercial flights to replace Dual-Pilot Operations (DPO) for next-generation flight operations, bolstered by ever-improving flight management systems (reliable support). However, it is crucial to acknowledge that human factors, such as workload, Situation Awareness (SA), decision-making, etc., are frequently discussed in aviation safety, which would be magnified in SPO. Current research on SPO is still nascent, and the existing literature is particularly prone to overlook human factors while thoroughly investigating the novel ground-/cockpit-based concepts for supporting SPO. This thesis endeavours to advance this area by developing a captain-centred monitoring system tailored for SPO, designed to promptly infer the single pilot’s incapacitation (undesirable states) through monitoring neurophysiological performance. This work will be the ‘Key’ to ‘Open’ the ground-/cockpit-based support for providing collaborative assistance to the single pilot, further forming a captain-centred, self-adaptive intelligence system, enhancing the safety and reliability of the upcoming SPO system from a human-centred perspective. The neuro-ergonomic and bio-sensor techniques will be employed to track the neurophysiological activity of the single pilot. Consequently, the neurophysiological pattern transitions of the captain operating from DPO to SPO would be comprehended, and then undesirable neurophysiological states directly/indirectly related to the potential pilot errors would be identified in real-time through the proposed three research stages (1) experimental investigation, (2) offline classification, and (3) online inference. Firstly, we conducted an exploration study to assess the feasibility of implementing neuro-ergonomic and bio-sensor measurements within the existing single-pilot system (general aviation cockpit) to realise the captain-centred monitoring system, which served as a prodromal exploration for our overall study. Through experimental investigation and offline classification modelling, the patterns of neurophysiological fluctuations were unveiled, and an offline classification model based on neurophysiological data was also developed to identify unsafe flight behaviour. By doing so, the feasibility of constructing the mentioned monitoring system within SPO using neuro physiological-based data was confirmed. After that, we turned to commercial aviation once the feasibility was confirmed. We empirically investigate how captains’ performance adapts during emergency scenarios when operating from DPO to SPO at first, employing a neurophysiological-based approach. These experimental-based noteworthy insights can inform commercial SPO measures, mitigating the persistent physiological fluctuations and assisting stakeholders in creating a captain-centred self-adaptive intelligence system to give captains adequate support. Secondly, we realised the CNN-based online inference to detect a single pilot’s low SA and associated mental workload in real-time using image-involved neurophysiological multimodal features. Upon completion of the three stages, the captain-centred monitoring system utilising neuro-ergonomic technologies was formed to predict single pilots’ incapacitation promptly and inform the ground-­/cockpit-based support, enhancing safety within the SPO aviation ecosystem.