Is this project an undergraduate, graduate, or faculty project?
Graduate
Project Type
individual
Campus
Worldwide
Authors' Class Standing
Eugene Pik, Graduate
Lead Presenter's Name
Eugene Pik
Lead Presenter's College
WW College of Aeronautics
Faculty Mentor Name
Emily Faulconer
Abstract
The 731% increase in the professional Unmanned Aerial Vehicle (UAV) fleet size, from 2016 to 2023, highlights the growing reliance on UAVs for commercial operations. This surge in usage has been accompanied by escalating safety concerns, highlighted by the National Transportation Safety Board’s (NTSB) investigations into UAV-related incidents. Addressing the gap in systematic categorization and visualization of UAV accident reports, this study leverages GPT-4's natural language processing capabilities and a suite of Python-based scripts and data analysis libraries. By analyzing 34 NTSB UAV accident reports, the research enhances the understanding of accident causality through the alignment of incidents with predefined NTSB categories. The study's findings identify system and component failures as the predominant causes of UAV mishaps. These accidents are widespread across the U.S., exhibit seasonal peaks, and have shown a notable reduction in frequency post-2019. This research underscores the effectiveness of AI and data visualization in dissecting UAV accident data, offering insights into trends and geographical distributions. In conclusion, it advocates for the integration of advanced machine learning and natural language processing techniques to refine UAV safety policies and highlights the anticipated role of AI in forecasting and mitigating future UAV accidents.
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
GPT-4 Assisted Categorization and Visualization of NTSB UAV Accident Reports
The 731% increase in the professional Unmanned Aerial Vehicle (UAV) fleet size, from 2016 to 2023, highlights the growing reliance on UAVs for commercial operations. This surge in usage has been accompanied by escalating safety concerns, highlighted by the National Transportation Safety Board’s (NTSB) investigations into UAV-related incidents. Addressing the gap in systematic categorization and visualization of UAV accident reports, this study leverages GPT-4's natural language processing capabilities and a suite of Python-based scripts and data analysis libraries. By analyzing 34 NTSB UAV accident reports, the research enhances the understanding of accident causality through the alignment of incidents with predefined NTSB categories. The study's findings identify system and component failures as the predominant causes of UAV mishaps. These accidents are widespread across the U.S., exhibit seasonal peaks, and have shown a notable reduction in frequency post-2019. This research underscores the effectiveness of AI and data visualization in dissecting UAV accident data, offering insights into trends and geographical distributions. In conclusion, it advocates for the integration of advanced machine learning and natural language processing techniques to refine UAV safety policies and highlights the anticipated role of AI in forecasting and mitigating future UAV accidents.