Changes in Response Patterns of Dismount Soldiers Due to the Reliability Level of Alert Systems 

The aim of this study was to examine how elements in system interface design affect response patterns and performance of a combat soldier when utilizing a display of information from unmanned vehicles. The tested factors were the influence of the reliability level of an alert system that was associated with the main display system, the importance of the presence of UAV video feed display in the user interface and the interaction between the two.

For achieving these goals, participants used a one screen display that included information in real-time from different remote information sources, video feeds derived from various unmanned vehicles that observed specific areas of interest or followed a particular vehicle. Participants were asked to perform detection, identification and classification tasks of suspicious vehicles or peoples in two different experiments. In addition, an alert system was available, on the same screen. It was designed as a separate system that receives information on events in real-time and is geared to help participants to attend to elements in the display.

The alert system is not perfect. The reliability level of the alert system depended on the rate of misses of true events and false alerts in irrelevant events. Four reliability levels were tested: high level, medium level, low level and without alert system (baseline group).

The experiments used a between- subjects design and took place in the department of Industrial Engineering and Management, at BGU. Participants were IDF reserve infantry soldiers.  The results were analyzed in three dimensions: 1. User performance; the rate of correct identification, wrong identification and response time of the user in cases of real event. 2. User performance using signal detection theory (SDT) and analysis of users` compliance and reliance by SDT measures as sensitivity level and response bias. 3. Users` eye scanning pattern of the display screen by dwell time in each region of interests (ROI) and the number of transitions between ROIs.

Overall, there was a clear effect for the presence of UAV video feed in the display on all three types of measures. Displays that included UAV lead to reduced rate of wrong identification, but also lead to reduction in the rate of correct identifications and longer response times. Respectively, analysis by SDT measures showed that the sensitivity level of the user increased in cases of not using UAV display and participants had a tendency to answer "yes". For the reliability level of the alert system, there was an influence mainly on users` scanning pattern of the screen. It was found that the low level reliability system caused shorter dwells, on one hand, and multi transitions between the ROIs, on the other hand.

In general, the results show meaningful and relevant connections between alert reliability and the content of the display with regard to the user response and user performance. There is a need to consider those factors when designing systems for combat soldiers and to take into account the possibility of changes in response and performance as a function of system design.