AI-Enabled UAV Systems for Disaster Response and Human Rescue: A Comprehensive Review

Abstract

The increasing frequency and severity of natural disasters, exacerbated by climate change, necessitate the development and deployment of advanced technological solutions for effective emergency response and mitigation. Unmanned Aerial Vehicles (UAVs), or drones, have emerged as a pivotal technology in this domain, offering unparalleled advantages in accessibility, mobility, and situational awareness. This comprehensive review synthesizes and critically analyzes the current state of research
in AI-enabled UAV systems specifically designed for disaster response and human rescue operations. We analyze four critical and interconnected domains: (1) the application of drone technology and computer vision in disaster relief, (2) advanced
multisensor technologies for robust human detection in challenging environments, (3) AI-enabled multimodal interaction systems for effective human-drone collaboration, and (4) comprehensive, end-to-end UAV-based disaster management frameworks. Our
integrated analysis reveals significant technological advances, particularly in detection accuracy (reaching up to 94.9% for human detection in visible spectra), sophisticated multimodal sensor fusion capabilities for all-weather operation, and the development
of autonomous systems for complex decision-making. However, persistent and significant challenges remain, especially in consistent night time operations, efficient energy management for extended missions, and navigating complex regulatory airspace
compliance. This review identifies these key technological gaps and systematically proposes future research directions, emphasizing the critical need for improved multimodal fusion algorithms, enhanced and diverse training datasets, standardized evaluation frameworks for objective comparison, and the development of integrated human-AI collaboration systems for next-generation,
resilient disaster response applications.