Workload and Reliability of Predictor Displays in Aircraft Traffic Avoidance

Abstract

In 2 experiments we describe the relevance of aircraft predictor information to the availability and deployment of visual attention. In both, airplane pilots fly a simulator in which flight path prediction is given bearing on the future state of their own aircraft and of a second traffic "intruder" aircraft that they must maneuver to avoid. The cockpit traffic display on which this information is depicted is an integral component of the concept of free flight or pilot self-separation. In Experiment 1 we show that added layers of predictive information improve performance, reduce mental workload (as subjectively measured), and that added complexity of the visual display thus resulting does not increase the inferred measure of head downtime (secondary-task performance). In Experiment 2 we examine the consequences to performance and visual attention if prediction is occasionally in error. We adopt the hypothesis that trust is related to the relative allocation of attention between the predictor symbol and the raw data of actual aircraft state. Such unreliability damages performance to some extent, particularly when the unreliable predictor forecasts more complex conflict geometry. This cost reveals the substantial allocation of attention to the predictor symbol. However, pilots, knowing the level of unreliability, appear to be relatively well calibrated in their allocation of attention between the 2 information sources.