Occlusion-Aware UAV-Based Monitoring in Complex Environments: Applications to Infrastructure and Surveillance

Andreas Valvis, Vasiliki Balaska, Antonios Gasteratos, Loukas Bampis

Keywords: UAV inspection, port monitoring, prompt guided segmentation, georegistration, object counting, critical infrastructure.

TL;DR: This paper presents an occlusion-aware UAV inspection framework for marine port monitoring that follows pathways, detects assets, revisits occluded viewpoints, and produces route-indexed inspection indicators.

Abstract:

Aerial inspection of complex infrastructure environments requires perception and control systems that can actively decide where to look next under uncertain and degraded visibility conditions. This paper presents an occlusion-aware UAV monitoring framework that integrates perception, navigation, and decision-making within a closed-loop inspection pipeline. The proposed approach couples prompt-guided pathway extraction, closed-loop route following, YOLOv11-based asset detection, overlap-driven occlusion reasoning, active viewpoint revisiting, and route-indexed counting within a unified inspection loop. The framework is primarily evaluated in the context of marine port
inspection in an Unreal Engine/AirSim environment, while its underlying perception-to-indicator logic is further demonstrated in additional use cases, including border traffic monitoring and mining site inspection.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 22