Autonomous Visual Inspection of Industrial Plants Using Unmanned Aerial Vehicles

The development of autonomous systems has spurred numerous innovative inspection strategies. Some operations, such as monitoring the condition of industrial structures, typically entail significant deployment of human resources and pose risks to human safety. In this context, this paper presents a visual inspection framework that leverages unmanned aerial vehicles to explore designated facilities, identifying structural damages such as cracks or fissures for inspection. The proposed approach integrates autonomous navigation and high-level decision-making capabilities to effectively explore predefined points of interest within partially known environments and to select and inspect candidate spots for further analysis. The framework is validated through both simulated and real-world experiments conducted in GPS-denied environments, utilizing only onboard UAV capabilities.