AI and Semantic Communication for Infrastructure Monitoring in 6G-Driven Drone Swarms
Tasnim Ahmed, Salimur Choudhury
TL;DR
The paper tackles real-time, automated infrastructure inspection with drone swarms by addressing 5G latency and reliability limitations and the high cost of manual inspections. It proposes a 6G-enabled framework that combines URLLC, semantic communication, edge AI, and LLM-based structured outputs to orchestrate a swarm and generate professional reports from semantic data. A conceptual system model, driving technologies, and a preliminary implementation are presented, including onboard road-monitoring AI and a network-performance analysis that highlights improvements in latency and reliability. The work demonstrates a path toward scalable, energy-efficient automated infrastructure monitoring and reports potential for digital twins to further enhance proactive management.
Abstract
The adoption of unmanned aerial vehicles to monitor critical infrastructure is gaining momentum in various industrial domains. Organizational imperatives drive this progression to minimize expenses, accelerate processes, and mitigate hazards faced by inspection personnel. However, traditional infrastructure monitoring systems face critical bottlenecks-5G networks lack the latency and reliability for large-scale drone coordination, while manual inspections remain costly and slow. We propose a 6G-enabled drone swarm system that integrates ultra-reliable, low-latency communications, edge AI, and semantic communication to automate inspections. By adopting LLMs for structured output and report generation, our framework is hypothesized to reduce inspection costs and improve fault detection speed compared to existing methods.