Digital Twin Framework Lets UAVs Inspect Pavement Without Lane Closures

Traditional pavement inspection often requires lane closures, causing traffic disruptions and safety risks. Researchers present a Unity-based digital twin framework for traffic-aware UAV pavement monitoring that operates without shutting down lanes. The framework integrates procedurally generated road defects, dynamic vehicles and pedestrians, autonomous UAV navigation, and a two-stage perception pipeline. First, a lightweight YOLOv8n detector localizes defects, vehicles, and pedestrians. Then a secondary classifier distinguishes between potholes, single cracks, and crocodile cracks. On the simulator test set, the pipeline achieved 99.26% overall accuracy across five classes, demonstrating high precision in a controlled environment.

The team evaluated three recovery strategies for occluded road segments: hover-and-recheck, micro-repositioning, and skip-and-revisit. Experiments varied traffic density and flight altitude, measuring coverage, mission time, energy consumption, and revisit ratio. Results show flight altitude strongly influences coverage, and adaptive recovery improves performance under occlusion. Hover-and-recheck delivered consistent coverage up to 97.03% under medium-to-high traffic, while skip-and-revisit excelled in low-traffic scenarios at 97.95% coverage. This work proves that digital twins can reliably simulate real-world conditions, allowing researchers and agencies to optimize UAV inspection routes and strategies before actual deployment—potentially reducing costs and road hazards.