Author Information

Is this project an undergraduate, graduate, or faculty project?

Undergraduate

Project Type

group

Campus

Daytona Beach

Authors' Class Standing

Gatlin Nelson, Junior

Lead Presenter's Name

Gatlin Nelson

Lead Presenter's College

DB College of Engineering

Faculty Mentor Name

Dr. Tianyu Yang

Abstract

Title: Dual-Perspective Risk Analysis for Human-LLM Decision Comparison in UAV Swarm Navigation   Unmanned aerial vehicle (UAV) swarms operating in low-altitude wireless network environments encounter localized disruptions that degrade positioning and navigation metrics. These disruptions are modeled as geographic failure zones with defined boundaries. A UAV discovers a zone by entering it and observing degraded performance on its onboard systems. This work assumes that affected UAVs can autonomously retreat to safety using onboard sensors and focuses on the subsequent rerouting decision. Once recovered, the system generates candidate repositioning points surrounding the vehicle, each scored using Conditional Value-at-Risk (CVaR). A human operator or a large language model (LLM) then selects which candidate to move the UAV toward as a new starting point for pathfinding. CVaR is computed from two concurrent perspectives: a ground-truth layer using actual zone geometry and an operator layer using only UAV-discovered estimates. The gap between them, alongside other measurements, is used to derive performance metrics at three scales: individual repositioning decisions, complete missions, and cross-level case studies. Six risk-based autonomy levels define thresholds that determine whether the human or the LLM makes each repositioning choice. This enables controlled comparison across a graduated spectrum from full human oversight to full LLM autonomy. The current work uses positional disruptions as a controlled starting point, but the framework is designed to generalize. Future work extends it to include SAE-inspired capability-based autonomy levels for domains such as cybersecurity monitoring and jamming countermeasures, where LLMs offer advantages over fixed algorithms by reasoning over ambiguous data. Preliminary automated testing confirms behaviorally distinct patterns across the six risk-based autonomy levels. Forthcoming human operator studies will compare repositioning decision quality against algorithmic and LLM baselines.

Did this research project receive funding support (Spark, SURF, Research Abroad, Student Internal Grants, Collaborative, Climbing, or Ignite Grants) from the Office of Undergraduate Research?

No

Share

COinS
 

CARS IMASS - Comparing LLM vs Human Operator Effectiveness In Multi-Agent Swarm Coordination

Title: Dual-Perspective Risk Analysis for Human-LLM Decision Comparison in UAV Swarm Navigation   Unmanned aerial vehicle (UAV) swarms operating in low-altitude wireless network environments encounter localized disruptions that degrade positioning and navigation metrics. These disruptions are modeled as geographic failure zones with defined boundaries. A UAV discovers a zone by entering it and observing degraded performance on its onboard systems. This work assumes that affected UAVs can autonomously retreat to safety using onboard sensors and focuses on the subsequent rerouting decision. Once recovered, the system generates candidate repositioning points surrounding the vehicle, each scored using Conditional Value-at-Risk (CVaR). A human operator or a large language model (LLM) then selects which candidate to move the UAV toward as a new starting point for pathfinding. CVaR is computed from two concurrent perspectives: a ground-truth layer using actual zone geometry and an operator layer using only UAV-discovered estimates. The gap between them, alongside other measurements, is used to derive performance metrics at three scales: individual repositioning decisions, complete missions, and cross-level case studies. Six risk-based autonomy levels define thresholds that determine whether the human or the LLM makes each repositioning choice. This enables controlled comparison across a graduated spectrum from full human oversight to full LLM autonomy. The current work uses positional disruptions as a controlled starting point, but the framework is designed to generalize. Future work extends it to include SAE-inspired capability-based autonomy levels for domains such as cybersecurity monitoring and jamming countermeasures, where LLMs offer advantages over fixed algorithms by reasoning over ambiguous data. Preliminary automated testing confirms behaviorally distinct patterns across the six risk-based autonomy levels. Forthcoming human operator studies will compare repositioning decision quality against algorithmic and LLM baselines.

 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.