Assessing the impact of low workload in supervisory control of networked unmanned vehicles

Abstract

This research investigated the effects of prolonged low workload on operator performance in the context of controlling a network of unmanned vehicles (UxVs) in a search, track, and destroy mission with the assistance of an automated planner. In addition, this research focused on assessing the physical, social, and cognitive coping mechanisms that operators rely upon during prolonged low workload missions. An experiment was conducted to collect data for researching the impact of low workload in human supervisory control of networked, heterogeneous UxVs. This research showed that performance was not necessarily affected at the low end of the workload spectrum, especially in the context of human supervisory control of networked UxVs. Given varying levels of low taskload, operators tended to gravitate toward a common total utilization (percent busy time) that was well above the required utilization. The boredom due to the low taskload environment caused operators to spend the majority of their time distracted; to a lesser degree, operators were more directed than divided in terms of attention. More directed attention predicted higher operator performance, especially in the tracking portion of the mission. Higher utilization predicted improved operator performance in search and destroy tasks, but hindered the automation's ability to track targets. Video gaming experience was a detriment to destroying hostile targets in this long duration, low workload mission involving human supervisory control of networked UxVs. Vigilance, shown by a decrement in amount of directed attention per hour, decreased over the course of the mission duration. Top performers had higher directed attention and coped with the boredom through extreme focus or use of switching times to stay engaged in the mission. In comparison to a moderate workload study, participants in this low workload experiment performed both better and worse. Low workload did not necessarily cause a drop in operator performance.