David B. Kaber

Effects of automation of information-processing functions on teamwork.

We investigated the effects of automation as applied to different stages of information processing on team performance in a complex decision-making task. Forty teams of 2 individuals performed a simulated Theater Defense Task. Four automation conditions were simulated with computer assistance applied to realistic combinations of information acquisition, information analysis, and decision selection functions across two levels of task difficulty. Multiple measures of team effectiveness and team coordination were used. Results indicated different forms of automation have different effects on teamwork. Compared with a baseline condition, an increase in automation of information acquisition led to an increase in the ratio of information transferred to information requested; an increase in automation of information analysis resulted in higher team coordination ratings; and automation of decision selection led to better team effectiveness under low levels of task difficulty but at the cost of higher workload. The results support the use of early and intermediate forms of automation related to acquisition and analysis of information in the design of team tasks. Decision-making automation may provide benefits in more limited contexts. Applications of this research include the design and evaluation of automation in team environments.

Investigation of multi-modal interface features for adaptive automation of a human-robot system

The objective of this research was to assess the effectiveness of using a multi-modal interface for adaptive automation (AA) of human control of a simulated telerobotic (remote-control, semi-autonomous robotic) system. We investigated the use of one or more sensory channels to cue dynamic control allocations to a human operator or computer, as part of AA, and to support operator system/situation awareness (SA) and performance. It was expected that complex auditory and visual cueing through system interfaces might address previously observed SA decrements due to unannounced or unexpected automation-state changes as part of adaptive system control. AA of the telerobot was based on a predetermined schedule of manual- and supervisory-control allocations occurring when operator workload changes were expected due to the stages of a teleoperation task. The task involved simulated underwater mine disposal and 32 participants were exposed to four types of cueing of task-phase and automation-state changes including icons, earcons, bi-modal (combined) cues and no cues at all. Fully automated control of the telerobot combined with human monitoring produced superior performance compared to completely manual system control and AA. Cueing, in general, led to better performance than none, but did not appear to completely eliminate temporary SA deficits due to changes in control and associated operator reorienting. Bi-modal cueing of dynamic automation-state changes was more supportive of SA than modal (single sensory channel) cueing. The use of icons and earcons appeared to produce no additional perceived workload in comparison no cueing. The results of this research may serve as an applicable guide for the design of human-computer interfaces for real telerobotic systems, including those used for military tactical operations, which support operator achievement and maintenance of SA and promote performance in using AA.

Comparison of Performance Effects of Adaptive Automation Applied to Various Stages of Human-Machine System Information Processing

Limitations in automation (expert system) capabilities and negative human performance consequences of automation in complex systems have led to the contention that use of computer assistance in high-level human-machine system information processing may be inappropriate. Adaptive automation (AA) has been explored as a solution to these problems; however, research has focused on the performance effects of dynamic control allocations of early sensory and information acquisition functions between human operators and computer controllers of complex systems. It has examined to a limited extent the human performance and workload effects of AA of cognitive tasks, such as decision-making, or of psychomotor functions such as response execution. This research compared the affects of AA applied to psychomotor tasks and cognitive tasks, including information monitoring, information analysis, decision-making, and action implementation, on overall human-machine system performance. Results demonstrated that operators are better able to adapt to AA when applied to lower level functions, such as information acquisition and action implementation, as compared to AA of information analysis and decision making tasks. The results also provided support for the use of AA, as compared to completely manual control.

Authority in Adaptive Automation Applied to Various Stages of Human-Machine System Information Processing

The goal of this study was to assess the performance and workload effects of applying adaptive automation (AA) to four stages of human-machine system information processing (information acquisition, information analysis, decision-making, and action implementation) and facilitating dynamic function allocations (DFAs) through two levels of computer authority (suggestion and mandate). The research was to provide insight into any interaction between these aspects of AA design. It was hypothesized that higher level automation, such as information analysis and decision making, would be more compatible with computer mandated allocations, while lower levels, such as information acquisition and action implementation, would be more effective under partial human control (computer suggestion and human veto). Results demonstrated that the effectiveness of AA is dependent upon both the type of automation presented to an operator and the type of invocation authority designed into the system. Performance with AA of information acquisition was superior to performance under decision automation. When using automated assistance, human acceptance of computer suggestions was superior to computer mandates. The results of this study may serve as an applicable guide for AA design in future complex systems.

The Influence of Individual Differences in Perceptual Performance on Pilot Perceptions of Head-Up Display Clutter

The objective of this study was to investigate the role of individual differences in pilot perceptual abilities on their experience of head-up display (HUD) clutter during a simulated instrument approach. Pilot contrast sensitivity, useful field of view (UFOV) and field dependence were assessed using standardized instruments. When these measures were included in a model of perceived clutter based on pilot subjective ratings of HUD configurations in terms of display clarity, contrast, density and similarity of elements, the predictive utility of the model modestly increased. Contrast sensitivity was found to be the most influential perceptual ability, with UFOV also being a predictor of perceived clutter. High contrast sensitivity for higher spatial frequencies was associated with higher ratings of overall clutter, while greater contrast sensitivity for moderate spatial frequencies was associated with lower ratings of clutter. Better performance on the UFOV measure of divided attention ability was associated with lower ratings of clutter, while better performance on the selective attention processing speed scale was related to higher clutter ratings. However, these variables had a much smaller degree of influence on clutter ratings than contrast sensitivity.

Situation Awareness in Driving While Using Adaptive Cruise Control and a Cell Phone

This research investigated the effects of an adaptive cruise control (ACC) system, and cell phone use in driving, on a direct objective measure of situation awareness (SA). Subjects drove a virtual car in a medium-fidelity driving simulation and performed a following task. Half of the subjects were required to respond to cell phone calls and all subjects completed trials with and without use of the ACC system. SA was measured using a simulation freeze technique and SA queries on the driving situation. Results indicated use of the ACC system to improve driving task SA under normal driving conditions, and cell phone conversations degraded SA. Results also revealed the ACC system to improve safe driving headway distance. Although the deviations in headway distance from an optimum were greater during cell phone conversations, this did not prove to be significant in terms of performance under normal driving conditions.

Human Performance with Vocal Cueing of Automation State Changes in an Adaptive System

Previous research on cueing of control mode changes in adaptive systems has focused on the use of visual cues with some work on complex auditory cues. Research has not explored the use of vocal cues for automation state changes and human performance implications. This study investigated vocal cues as feedback on adaptive robotic system states for supporting operator performance when responding to control mode transitions compared with visual (icons) and auditory (earcons) feedback. Thirty-two participants performed a virtual reality simulation of a telerobot-assisted underwater mine disposal task. Modal cues were presented with task phase changes and robot control mode changes. Cue type and complexity (length of messages) were varied between and within subjects, respectively. Operator performance was evaluated in terms of time-to-task completion, workload, and situation awareness. Results demonstrated vocal cues to be superior to simple visual icons and no cueing for performance. Earcons did not produce worse performance ratings than vocal cues for complex messages. The results of this study are applicable to the design of future automated systems.

Design of a Cognitive Model-Based Decision Support Tool for Anesthesiology Crisis Management

A decision support tool was developed for use by anesthesia providers in crisis situations. The tool alerts anesthetists to a developing crisis, manifested by changes in patient physiological variables, and provides them with a list of preventive measures for dealing with the crisis. A novel approach to the design of the tool was defined, including: (1) performing a hierarchical task analysis to identify anesthetist procedures in detecting, diagnosing and treating a critical incident; (2) carrying out a cognitive task analysis to elicit goals, decisions, and information requirements of anesthetists during crisis management procedures; (3) coding of natural language information recorded in the task analyses in a computational GOMS (goals, operators, methods, selection rules) cognitive model; and (4) prototyping an interface to present output from the cognitive model using ecological interface design principles. A preliminary validation of the tool and interface was performed with anesthesiology and usability experts.

Bridging the gap between Human — Automation Interface Analysis and Flight Deck Design Guidance

Next generation aviation operations will place a much greater dependence on automation usage, and therefore additional emphasis needs to be placed on the evaluation of human automation interaction in the design and evaluation of these systems. Additionally, new airworthiness regulations and regulatory certification processes are beginning to focus on the design and verification testing of the pilot-automation interaction. Current human computer interaction analyses (computational human performance models and task analysis methods) are not effectively usable within the constrained timeline of real world design and certification processes. Fundamental and theoretical work is needed to develop methods and tools that will provide designers and regulators with the means of testing and providing useful feedback about the efficacy of these interactions.