Operator state monitoring for workload prediction and management

Abstract

Ongoing efforts to modernize the U.S. Army is resulting in the development of the next generation of rotary aircraft under the Future Vertical Lift (FVL) program. FVL missions will be characterized by increased agility, degraded visual environments (DVE) and optionally piloted vehicles (OPVs) in complex, highly contested and dynamically changing environments. New technologies and automation in the cockpit allow aviators to be more effective in completing their missions in the modern dynamic battlefield. Autonomous systems have the potential to contribute to improvements in operational safety, efficiency and effectiveness but have also introduced additional human factors engineering considerations. These technologies provide pilots with appropriate and timely information and support, while avoiding overloading with excessive clutter and information. At the same time, excessive automation can lead to overloading/ underloading, leading to automation misuse, complacency, and loss of situational awareness (SA). Adaptive decision aiding systems have the potential to enhance operators’ capabilities by recognizing situations, external (environment) and internal (operator state) and providing subsequently real-time adaptation of the aircraft controls and human/ machine interfaces. A pre-requisite to the development of such adaptive decision aiding systems is the real-time monitoring and identification of the operator’s performance and mental state in terms of workload and situation awareness. Being multidimensional, latent factors, workload (WL) and SA need to be inferred through observable variables such as subjective assessments, task-based metrics, and psychophysiological measures. Of particular interest for FVL are techniques and technologies that enable real-time WL and SA assessment, for realistic missions in simulator and/or real flight conditions. The present research aims to provide a methodology to manipulate WL and SA in an operationally relevant mission in a fixed-platform UH60 simulator, evaluate the usability, diagnosticity, sensitivity and reliability of various performance metrics in situ using novel approaches for the data analysis and validate real-time predictive models of WL. An experiment was designed to manipulate WL and SA by using two levels of flight control level of automation (LOA) and two levels of obstacle cueing symbology during a degraded visual environment (DVE) medical evacuation (MEDEVAC) mission. The level of WL was further manipulated by the presence of obstacles and aircraft survivability equipment (ASE) threat events (radar and missile warnings). Five UH60M Army experimental test pilots participated in the simulation at the U.S. Army rotorcraft in flight laboratory RIFL system integration laboratory (SIL) at Fort Eustis, VA. Preliminary analyses validate the WL driver’s selection, and the usability of the real-time subjective WL report using a modified Bedford rating scale, used to compute the spare capacity operations estimator (SCOPE). A novel approach to statistical analyses is proposed, that will allow to determine the relative weight of each metric to the final WL and SA estimates, and support performance modeling.