Abstract

In this thesis, we study the problem of optimal search and coverage with heterogeneous team of unmanned aerial vehicles (UAVs). The team must complete the coverage of a given region while minimizing the required time and fuel for performing the mission. Since the UAVs have different characteristics one needs to equalize the ratio of the task to the capabilities of each agent to obtain an optimal solution. A multi-objective task assignment framework is developed for finding the best group of agents that by assigning the optimal tasks would carry out the mission with minimum total cost.

Once the optimal tasks for UAVs are obtained, the coverage area (map) is partitioned according to the results of the multi-objective task assignment. The strategy is to partition the coverage area into separate regions so that each agent is responsible for performing the surveillance of its particular region. The decentralized power diagram algorithm is used to partition the map according to the results of the task assignment phase. Furthermore, a framework for solving the task assignment problem and the coverage area partitioning problem in parallel is proposed. A criterion for achieving the minimum number of turns in covering a region a with single UAV is studied for choosing the proper path direction for each UAV. This criterion is extended to develop a method for partitioning the coverage area among multiple UAVs that minimizes the number of turns.

We determine the "best" team for performing the coverage mission and we find the optimal workload for each agent that is involved in the mission through a multi-objective optimization procedure. The search area is then partitioned into disjoint subregions, and each agent is assigned to a region having an optimum area resulting in the minimum cost for the entire surveillance mission.