The UAV-Mission Planning Problem

In military operations Unmanned Aerial Vehicles (UAVs) are used for reconnaissance of target locations in the area of operations. These target locations each have their own priority. In this thesis the UAV-Mission Planning Problem (UAV-MPP) is addressed where the fuel usage of the flight between each pair of targets is known a priori only probabilistically and the information about the target locations can only be obtained within their assigned time window. The goal is to maximize the total gathered information value during a flight which is restricted by the fuel capacity of the UAV. This problem can be modeled by the Stochastic Orienteering Problem with Time Windows (SOPTW), which is both practically and theoretically relevant. To solve this problem, two different approaches are presented. The first approach constructs an initial tour before the flight which is adjusted to the realized fuel usages during the flight. To construct an initial tour, different stochastic programming models are used including three variants of a chance constrained programming model and two variants of a recourse model. The second approach is a One-Step-Ahead Routing (OSAR) approach in which the next location is determined after recording the previous. A case study is performed to evaluate and compare the presented methods.