The use of micro unmanned aerial vehicles (UAVs) is gaining popularity in various areas like disaster management and construction site monitoring due to their ease of use and low costs. In this thesis we consider the exploration of a disaster scene, e.g., earthquake or train accident, using micro UAVs equipped with cameras. The terrain is represented as a set of locations to be observed with potentially different priorities. The aim is to always have the most up-to-date information for every location. For this purpose, fast automatic planning is required, as human resources cannot be wasted on steering the drones.
In order to model the described task, we introduced the continuous monitoring problem and its extensions with inter-depot routes and priorities. Continuous monitoring means that a fleet of vehicles has to periodically visit a set of locations. Vehicles are constrained by the capacity of their energy storage that can be renewed at one of the base stations. The goal is to find a plan where the fleet visits locations uniformly and as often as possible but avoids long delays between revisits.
To solve these problems, we propose several construction approaches:
modified Clarke and Wright algorithm, queue-based insertion heuristic, inter-depot insertion heuristic (IDIH), and IDIH with reservations. The solutions obtained by these algorithms are improved by a metaheuristic based on the variable neighborhood search. The suggested methods were evaluated on numerous instances including real-life scenarios. The evaluation showed that the developed methods are capable of providing near-optimal solutions in a short time.