Algorithmic and combinatorial problems on multi-UAV systems

Abstract

Mathematics has always been a fundamental piece in robotics and, research in robotics has played an important role in the development of mathematics. This thesis is motivated by the growing interest on problems that appear in aerial robotics applications, specifically, on cooperative systems of multiple aerial robots or drones. Most of the research works in multi-robot systems have focused primarily on construction and validation of working systems, rather than more general and formal analysis of problems and solutions. By contrast, this thesis focuses on formally solving problems of aerial multi-robot systems from a discrete and combinatorial optimization perspective. Inspired on problems of this area, the thesis introduces some new theoretical models and problems of interest for mathematicians and computer scientists. The following topics are covered in this thesis: (1) synchronization: design of a coordination strategy to allow periodical communication between the members of a cooperative team while performing a task along fixed trajectories in a scenario with limited communication range, (2) robustness: analysis of the detrimental effects in the performance of a synchronized system when one or more robots fail, (3) stochastic strategies: performance analysis of a synchronized system using drones with stochastic decision making, and (4) task allocation: decentralized coordination to perform periodical task allocation in order to maintain a balanced work load for all members of a team with limited communication range. In the first part of the thesis, we study the synchronization problem giving a theoretical characterization of the solutions and, we present an algorithm to build a synchronized system for a given set of covering trajectories. The second part focuses on the study of the robustness in a synchronized system regarding to two key aspects: covering of the working area and communication between the members of the team. We rigorously study several combinatorial problems to measure how robust a system is to deal with drones failures. Connections of theseproblemswithnumbertheory, graphtheory, circulantgraphsandpolynomial multiplication are shown. The third part is devoted to an analysis of synchronized systems using random aerial robots. This topic is closely related to the random walk theory. It is shown that stochastic strategies increase the robustness of a synchronized system. Finally, this thesis introduces the block sharing strategy to addresstheproblemofmaintainingabalancedtaskallocationamongtherobotsby using periodical communications. A proof on the convergence to an optimal task allocation is given and, a case study for structure construction using a cooperative team of aerial robots is presented. All algorithms developed in this thesis have been implemented and extensive experiments have been conducted to analyze and validate the proposed methods.