Changes in technology, particularly for example in commercial air operations, have led to incremental increases in the number and variety of associated safety procedures and checklists. This work addresses concerns about how people and systems respond to unanticipated events, as applicable to such commercial air operations, and to examine whether these “if–then” approaches prove sufficient to respond to prospective operational uncertainties. The current work draws from the literature on systems resilience, cognitive flexibility, and adaptation to consider how response strategies that are integrated into human–machine training at all levels of operation can aid in effective resolution to such unanticipated events. A number of scientific insights, methods, and domains are identified as being able to be employed to avoid catastrophic failure in current and prospective operational environments. While heuristics for advisement do provide an initial level of defensive protection, evolving airspace operations need to be adaptive to, and resilient in respect of, emerging and even unanticipated challenges. Prospective response strategies need to encompass both the demands that can be evidently foreseen and those that remain at present, indeterminate. Resilience in responding appears to be a primary dimension of success in relation to these challenges. The information herein distilled can increase operator performance and aviation systems’ response to nonproceduralized and unanticipated events as well as being applied to a vast array of other safety-critical operations beyond this one realm.

Few today would dispute that the age of autonomous machines is nearly upon us (cf., Kurzweil, 2005; Moravec, 1988), if it is not already. While it is doubtful that one can identify any fully autono...

To begin, I would first like to thank those who were kind enough to comment upon my paper, (Hancock, 2021) and even those whose comments were not so kind. For, after all, it is only by postulation,...

A detailed understanding of operator individual differences can serve as a foundation for developing a critical window on effective, adaptable, user-centered automation, and even for more autonomous systems. Adaptable automation that functions according to such principles and parameters has many potential benefits in increasing operator trust and acceptance of the automated system. Our current study provides an assessment of the way that individual differences in attentional control (AC) affect the preference for a selection of a desired level of automation (LOA). Participants who scored low or high on AC were either allowed to choose among four possible LOAs or restricted to a predetermined LOA. These manipulations were engaged while the operator was performing visual and auditory target detection tasks. The AC level was found to be inversely proportional to the LOA preference. Operators also performed better when they were preassigned to a fixed LOA rather than given a choice. Individual differences can thus be shown to affect the performance with the automated systems and should be considered in associated design processes. When deciding whether to give the operator control over LOA in a complex system, engineers should consider that the amount of control that operators may want does not necessarily reflect their actual needs.

Our work posits that existing human-animal teams can serve as an analog for developing effective human-robot teams. Existing knowledge of human-animal partnerships can be readily applied to the HRI domain to foster accurate mental models and appropriately calibrated trust in future human-robot teams. Human-animal relationships are examined in terms of the benefiting roles animals can play in enabling effective teaming, as well as the level of team interdependency and team communication, with the goal of developing applications in future human-robot teams.

In all human-robot interaction, trust is an important element to consider because the presence or absence of trust certainly impacts the ultimate outcome of that interaction. Limited research exists that delineates the development and maintenance of this trust in various operational contexts. Our own prior research has investigated theoretical and empirically supported antecedents of human-robot trust. Here, we describe progress to date relating to the development of a comprehensive human-robot trust model based on our ongoing program of research.

Augmented reality (AR) entails overlaying the real world with information from computer-generated displays. Current AR technologies support limited mobility, although this is expected to change in the future. This paper presents experimental results of effects of various AR display strategies on human performance in a simulation-based analog of a ldquosearch and rescuerdquo navigation task. The augmentation scheme was a spatially and temporally registered map that was overlaid onto a corresponding real-world maze. The experiment required the participants to traverse the maze, periodically answer orientation questions, obtain a target object, and exit the maze as quickly as possible. One hundred twenty participants were evaluated in six different conditions. There were two control conditions (paper map or compass prior to entering the maze), and four experimental conditions (combinations of egocentric and exocentric maps, and continuously on or on-demand map display). Performance measures consisted of duration of time to traverse the maze and percentage of maze covered. AR resulted in better performance than the control conditions in terms of accuracy by facilitating the participants' coverage of the maze. Results show that the better performance with respect to time was in the map control condition. This result may be due to the small size of the maze, which could be memorized. However, AR is expected to exhibit better performance compared to a paper map, when more complex environments are employed. These results demonstrate promising benefits in mobile AR usage in specific navigation tasks. Design guidelines were extracted to guide future AR systems continued progress in enhancing performance.

In this paper, the responses of experienced professional pilots to change in interface configuration and differing automated invocation procedures were examined using a simulated flight-task environment. Performance was evaluated on three subtasks: two-dimensional compensatory tracking; fuel management; and systems monitoring. The status of automation, which was available for tracking and fuel management only, was conveyed by a change in display configuration represented by either a reduction in the size of the relevant display or by a reduction in size accompanied by its displacement to a peripheral spatial location. Combined with these interface-configuration changes were two forms of automation invocation procedure, which were pilot-initiated automation and system-initiated automation. Each was compared to a standard manual-control condition. Results indicated several response asymmetries. While tracking showed no effect for the location of the automated fuel-management display, fuel-management performance did reveal a significant effect, which favored the peripheral location of the automated tracking display. This display-location effect is thought to result from a general requirement for pilots to change their visual-scan pattern. The converse effect does not appear for fuel management and represents the continued primacy given by the pilots to tracking performance. System-initiated automation of fuel management, as a set condition, resulted in significantly better tracking performance, in both mean and variability measures, when compared to pilot-initiated automation. In the converse situation, involving the automation of the tracking subtask, a significant difference was also evident, but only in the variance measure of the fuel-management performance. The fuel-management variance for the pilot-initiated automation of tracking was significantly lower than that for the condition where the automation was enacted by the system. These results indicate that the automation-initiation process itself influences subsequent multitask performance. The present results support a general contention that the operator should initiate automation, except in circumstances in which the operator is for some reason incapacitated

This paper examined the effects of different forms of automation invocation, the reconfiguration of the task display when automated, and the influence of the specific modality that warned of the manual/automation transition on operator performance. Thirty-two experienced pilots engaged in a multiple-task situation consisting of tracking, monitoring, and fuel management subtasks, representative of typical aviation demands. Automation of the tracking task could be invoked in four different ways: 1) system-initiated automation (SIA); 2) pilot command by negation (PCN); 3) pilot command by initiation (PCI); and 4) pilot-initiated automation (PIA). Pilots were warned of the mode change between manual and automated controls by either a visual, an aural, or a combined visual and aural cue. The display of the subtask while in automation was reduced in size and placed either in a central or peripheral location. Results indicated that SIA had a differential effect on tracking performance as compared to all other forms of automation invocation. The respective location of the automated display had it effects in the fuel management subtask, whereas monitoring capability remained stable across manipulations. A significant three-way interaction between invocation procedure, display location, and warning modality illustrated the selective disadvantage of the visual warning combined with the central location under the PCI procedure on tracking response. Measures of subjective response suggested that visual warning modality was slightly more taxing than either an auditory or a combined auditory and visual warning. Pilots also experienced elevated fatigue when the system initiated the automation. These results confirm that both performance and subjective perception of multitask demand are greater when the system controls the option to automate. A qualitative model is presented, which provides an approach for the integrated assessment of human performance with adaptive systems.

Three experiments examined the use of vibrotactile cues to guide an operator toward a target. Vibrotactile stimulation on the hand can provide spatially stabilizing cues for feedback of subtle changes in position. When such feedback is present, a deviation from the point of origin results in tactile stimulation indicating the direction and magnitude of the positional error. Likewise, spatial deviation from a desired position displayed tactually can provide robust position guidance and stabilization sufficient to improve the acquisition time and accuracy of fine cursor control. A major advantage of this mode of information representation is that it can be present at the same time as visual cues with minimal cross-modal interference. Our findings suggest that performance is actually enhanced when both tactile and visual cues are present. Although previous studies have suggested that various forms of tactile feedback can provide position guidance and stabilization, to our knowledge, this work is the first that details the effect of tactile feedback on target acquisition directly.

It is argued that the information processing model represents an incomplete foundation for a human-factors account of cognitive tasks. The authors propose the inclusion of purposive and proactive accounts that are more consistent with the requirements for explaining goal-oriented reasoning and behavior. They review a number of related approaches and sketch out the beginnings of a theoretical foundation for human factors in the form of assumptions that bound human behavior. The initial focus is on basic assumptions governing the availability and use of cognitive resources, but the assumptions suggested elsewhere (e.g. the model human processor approach) are also utilized.>

In examining the role of time in mental workload, the authors present a different perspective from which to view the problem of assessment. Mental workload is plotted in three dimensions, whose axes represent effective time for action, perceived distance from desired goal state, level of effort required to achieve the time-constrained goal. This representation allows the generation of isodynamic workload contours that incorporate the factors of operator skill and equifinality of effort. An adaptive interface for dynamic task reallocation is described that uses this form of assessment to reconcile the joint aims of stable operator loading and acceptable primary task performance by the total system.>