Use this URL to cite or link to this record in EThOS:
Title: Verifiable swarm engineering with limited communication
Author: Punzo, Giuliano
Awarding Body: University of Strathclyde
Current Institution: University of Strathclyde
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
The new paradigms of swarm engineering, distributed architectures and autonomous multi-agent systems, are foreseen to redefine the way many engineering problems are approached. The affirmation of these new concepts requires the complete understanding of complex dynamics by the designers. That is, any system whose concept departs from the monolithic architecture must deliver its tasks in a predictable way and be controlled in a safe manner, while keeping the maximum possible autonomy. This work aims to span the gap between a complete foreseeable behaviour and system autonomy using precise mathematical descriptions of the dynamics and control of multi-agent systems. Dynamical system theory, Lyapunov stability, linear algebra and graph theory are used to rigorously frame the problem and delineate the characteristics of such systems in relation to a number of applications and performance parameters. The work first considers multi-agent systems as multi-particle systems in a physics fashion to draw fundamental results about the robustness to fragmentation when the individuals do not benefit from all-to-all communications. The exploitation of limited communications together with artificial potential functions is shown to be an effective way to shape formations of agents in a range of applications for future engineering, and in particular this scheme is proved to be effective for space-based communications through the autonomous deployment of antenna arrays. In this context, application to robotics is explored through laboratory tests exploiting wheeled robots with possible applications to structural inspection or planetary exploration. A stable fractal formation is proved to emerge out of a number of agents whose interaction network presents a recursive layout whereby relative motion is driven by artificial potential functions. Finally, the fast manoeuvring problem is covered together with one of allocating resources in an efficient way to track an external signal for the benefit of the group as a whole. Through an algebraic approach, the tracking capabilities are distributed amongst the agents producing advantages at group level for the tracking of an external signal. This also translates into fast reaction to threats.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral