Situation Awareness-Based Agent Transparency and Team Fluency in a Multi-Agent Intelligence, Surveillance, Reconnaissance Scenario

Future intelligence, surveillance, and reconnaissance (ISR) missions are likely to feature human operators supervising multiple autonomous aircraft rather than directly controlling a single platform. In these multi-agent settings, operators must rapidly infer what autonomous teammates are doing, why they are doing it, and what they are likely to do next. Situation Awareness-based Agent Transparency (SAT) is a principled approach that maps agent disclosures to Endsley’s three levels of situation awareness (SA). This paper investigates how cumulative SAT levels affect human-autonomy team fluency in a multi-agent ISR (MAISR) game where a human-controlled and an agent-controlled aircraft collaborate to classify targets under time and health constraints. Twenty-four experienced gamers completed four MAISR missions under a within-subjects design with four levels of SAT. Dependent measures included mission performance, behavioral indicators of reliance and team fluency, NASA-Task Load Index workload ratings, trust in the agent, and mid-mission SA probes using the Situation Awareness Global Assessment Technique (SAGAT). Repeated-measures analyses showed that cumulative transparency significantly improved mission performance and encouraged greater reliance on the autonomous teammate, with the largest gains observed under full transparency. Workload tended to decrease with higher transparency, whereas trust and SAGAT-based SA did not differ reliably across conditions. These findings suggest that SAT-style robot-to-human transparency primarily shapes how human-autonomy teams coordinate and allocate work rather than directly improving mid-mission SA.