The Complex Dynamics of Team Situation Awareness in Human-Autonomy Teaming

This study explores how Human-Autonomy Teams work together in a Remotely Piloted Aircraft System (RPAS) to overcome three types of degraded conditions, including automation and autonomy failures, and malicious attack. The two human participants were informed that the pilot was a "synthetic" agent that has limited communication capacity. For in-depth exploratory analysis, we identified one high- and one low-performing team in terms of overcoming failures and malicious attack, and then we used nonlinear dynamical methods to understand how human-autonomy interactions might affect overall Team Situation Awareness (TSA) in terms of level of complexity. We first produced Joint Recurrence Plots (JRP) to demonstrate predictability of team communication behavior during the TSA. After that, in order to identify how flexible the team was during degraded conditions, we examined entropy across four layers to represent RPAS:communication - chat-based interactions; vehicle - the RPA itself; control - user interface; and system - total activity of all layers. Results from the JRP showed that the high performing team communicated more effectively than the low performing team during the all three types of failures, while the entropy analysis showed that the high performing team appeared to be more flexible in their communication and overall system patterns.