Monitoring Pilot Driving Status based on Facial Features and Physiological Signals under a Single-pilot Flight Mode

Abstract

This paper systematically reviews pilot state monitoring technologies based on facial features and physiological signals under the Single Pilot Operations (SPO) mode. The SPO mode replaces the traditional two-pilot interpersonal coordination with a ‘human-machine-ground’ collaboration, which can reduce crew costs and fuel consumption, but relies on technical monitoring to fill the gap of ‘interpersonal supervision.’ The study summarizes the core principles and indicators of facial feature monitoring and physiological signal monitoring, analyzes the performance of multimodal data fusion methods, machine learning, and deep learning models in state recognition, and validates the role of multimodal fusion in improving recognition accuracy through related research findings. Current technologies face challenges such as facial occlusion, insufficient comfort in physiological signal acquisition, lack of data privacy, poor model generalization, and inconsistent industry standards. Future efforts should optimize device design and strengthen interdisciplinary collaboration to promote standardization, thereby facilitating the practical application and adoption of SPO monitoring technologies.