SELF-STABILIZED GROUPS OF MICRO AERIAL VEHICLES

Abstract

The goal of this habilitation thesis is to aggregate our research on stabilization and con- trol of cooperating robots in real environments without the need to build a dedicated

infrastructure. The thesis presents methodologies and enabling techniques for controlling teams of autonomous robots acting in a shared working space, using onboard localization sensors only. The main contribution of this work lies in the design of methods for motion planning

and stabilization of formations of ground and aerial robots, and for bio-inspired behav- ioral control patterns for swarms of multi-rotor micro aerial vehicles (MAVs). All pre- sented approaches are designed to satisfy the constraints imposed by a system of onboard

relative localization of teammates, which was designed as the main sensor to enable the

deployment of groups of cooperating robots in environments without external localiza- tion. The attached publications present theoretical studies specifying the requirements for

the relative localization sensor to satisfy the stability of MAV swarms and convergence of motion of compact formations. In addition to the theoretical contribution, a large part of this document is dedicated to

applications of the methodologies, since robotics is an application-oriented and application- motivated scientific field. Most of the methods presented here have been verified experi- mentally in environments reflecting the target applications. Continuous experimental ver- ification of the methods is an added value of the results that have been achieved. Direct

interconnection between the methods designed for MAV group control and stabilization and the properties of the MAV sensors described in the real-world experiments is a crucial attribute of the work presented here. It has enabled the scientific results to be transferred into real deployment and into industrial applications, as has been shown in numerous

examples in the thesis. Intensive validation of system performance in demanding envi- ronments was also a key factor in the MBZIRC 2017 competition, where our solution for

a multi-robot challenge outperformed the results of all other competitors, most of whom had relied on laboratory testing only. Cohesive linkage between theoretical principles, on the one hand, and real experience and industrial solutions, on the other, has also provided a decisive competitive advantage for students of our team in their further career.