Augmenting task-centered design with operator state assessment technologies

• Authors: Karl F. Van Orden;Erik Viirre;David A. Kobus
• Venue: FAC'07 Proceedings of the 3rd international conference on Foundations of augmented cognition
• Year: 2007

Task-Centered Design (TCD) of human-system interfaces focuses on supporting the user throughout all phases of tasks, from initiation to completion. TCD typically requires software that monitors aspects of system information to trigger tasks, develop user-friendly information sets, propose task solutions and actions, and confirm actions as directed and approved by the operator. The operator monitors tasks awaiting completion on a Task Manager display. We demonstrate that moment-to-moment operator workload monitoring is greatly facilitated by TCD. Workload estimates were obtained every 2-min over the course of a 35-min test session during an air defense command and control scenario. Workload was readily modeled by the task loading, and the density of track icons on the display. A second study related the unitary workload estimates to NASA TLX workload subscales. Unpublished data from our laboratory indicated that eye activity measures (e.g., blink frequency and duration, pupil diameter, fixation frequency and dwell time) did not improve the estimation of workload. These findings indicate that at least for well-executed TCD systems, eye tracking technologies may be best employed to monitor for fatigue and incongruities between the focus of attention and task requirements. Recent findings using EEG hold promise for the identification of specific brain signatures of confusion, orientation, and loss of situational awareness. Thus the critical element of human directed systems is good initial design. Understanding of the task will lead to system automation that can balance the workload of the operator, who is functioning in a normal state. However, physiological monitoring will be most useful if operators veer beyond their normal conditions and are confused, overloaded, disoriented or have other impairments to their abilities. By detecting the operator's loss of function early, inappropriate operator inputs can potentially be avoided.