Date of Award
12-2025
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Industrial Engineering
Committee Chair/Advisor
Dr. Kapil Chalil Madathil
Committee Member
Dr. Sudeep Hegde
Committee Member
Dr. Nathan McNeese
Committee Member
Dr. Patrick Rosopa
Abstract
Military teams frequently operate in hazardous and dynamic environments where they must recognize threats, act swiftly, and maintain awareness of their surroundings. To assist soldiers in identifying threats, analyzing data, and creating safer routes, Artificial Intelligence (AI) systems are being integrated into these teams. But only when humans comprehend what AI is doing, why it is doing it, and how to collaborate with it can AI be helpful in these high-stakes missions. To enhance human-AI collaboration during military route reconnaissance missions, this dissertation explores ways to improve human-AI communication. In a route reconnaissance mission, teams of soldiers must navigate through unfamiliar terrain, identify potential threats like enemy forces or roadside bombs, and determine the best course of action. The environmental conditions change rapidly, and the information obtained and exchanged during the mission may be ambiguous or incomplete, making this task challenging. This dissertation's research focuses on how AI teammates should communicate with soldiers to enhance team performance, maintain situational awareness, and prevent information overload. The first part of the dissertation reviews past studies on how people and autonomous systems communicate when working together as a team. The review suggests that incorporating multiple types of communication, such as combining video, sound, or vibration, can facilitate a better understanding of AI. However, these benefits have rarely been tested in military settings. This gap underscores the importance of understanding how multimodal communication can improve the performance of human-AI teams in real-world missions. The second part of the dissertation reports findings from interviews with eighteen active-duty soldiers who have experience conducting route reconnaissance operations. These soldiers stated that they would trust AI teammates more if the AI communicated clearly, explained its decisions, and provided timely updates during the mission. They emphasized that AI should support them by providing mission-critical information that helps them make decisions during time-critical moments. These insights were used to design a virtual reality simulation that closely mirrors real reconnaissance missions. The third study examined the impact of two key AI features, transparency (showing what the AI is doing) and explainability (explaining why it makes recommendations), on team performance in human-AI teams. Participants collaborated with AI teammates at varying levels of transparency and explainability. The results showed that when AI teammates had explainability, people detected more threats, missed fewer dangers, and experienced a lower workload. This means that when humans understand the AI’s actions and its reasoning, they work more effectively as a team. The final study examined how different ways of communicating, such as visual displays, audio messages, or a combination of both, affect teamwork. The results showed that adding audio to the AI’s visual updates helped people maintain better situational awareness, complete tasks faster, and feel less time pressure. Audio cues enabled participants to maintain their focus on the environment while still receiving important information from AI. The study also found that allowing humans to communicate with each other through voice helped them coordinate more easily. Overall, this dissertation demonstrates that the design of communication between humans and AI has a significant impact on their joint performance in military operations. Clear, timely, and multimodal communication through audio and visual channels helps humans stay aware of threats, understand the behavior of AI teammates, and make better decisions under pressure. These findings offer guidance for designing future AI systems that operate safely and effectively alongside soldiers in complex and hazardous environments.
Recommended Citation
Sawant, Sarvesh, "Interfaces to Enhance Performance in Human-AI Teams Conducting Safety-Critical Military Operations: A Route Reconnaissance Use Case" (2025). All Dissertations. 4121.
https://open.clemson.edu/all_dissertations/4121
Author ORCID Identifier
https://orcid.org/0009-0001-8508-0171