Solving Infrastructure Monitoring Problems with Multiple Heterogeneous Unmanned Aerial Vehicles

A number of critical infrastructures, such as gas or oil pipelines operate in a sensitive environment and any damage done to the infrastructure significantly harms the surrounding fauna and flora and poisons water supplies. The infrastructure thus needs to be monitored and any damage or failure has to be detected and reported as quickly as possible. We focus on the enhancement of the monitoring systems— we propose a first holistic solution for a set of heterogeneous unmanned aerial vehicles (UAVs) which monitor the infrastructure under current technological restrictions such as speed, battery endurance and sensing radius. We solve the problem of (1) the allocation of charging/maintenance stations in the area, (2) the assignment of the UAVs to the stations and (3) the computation of their trajectories with respect to the environment sensitivity. We propose a formal graph-based model capturing problem constraints and requirements. We explore possible decompositions of the problem and we propose a number of algorithms allowing to choose between algorithm runtime and solution quality. The results show that our approach can be used to monitor realworld sized infrastructures of a length of tens of kilometers using up to five UAVs.