Swarm Control for Mass UAS Developments in Disconnected Environments

Simple item page
dc.contributor.advisorSzabo, Claudia
dc.contributor.advisorCoyle, Andrew
dc.contributor.advisorHunjet, Robert
dc.contributor.authorFraser, Bradley Robert
dc.contributor.schoolSchool of Computer Scienceen
dc.date.issued2022
dc.description.abstractUninhabited Aerial Systems (UASs) are one of the best examples of recent disruptive technology and have justifiably become the workhorses for myriad novel military and civilian applications. UAS technology has advanced to the point where the deployment of swarms of UASs is a reasonably viable activity from a logistics point-of-view. However, fully autonomous and distributed control of such swarms remains elusive. In particular, the case where swarm members, or other network nodes they are supporting, exist in a disconnected communications state makes swarm coordination especially difficult. The high-stakes nature and potentially dangerous consequences of such research and development activities also makes their implementation exceedingly rare. Furthermore, issues pertaining to scalability of algorithms from the perspective of automated design and deployment still exist. This thesis aims to address these problems through simulated and robotic field experiments, using bio-inspired and reinforcement learning approaches to generate swarm control schemes for common UAS applications. Of the technical parts of the thesis, Chapters 4 to 6 propose several novel swarm control algorithms to support communications and other positionalbased missions. Mathematical analysis of the resultant emergent behaviour provides insight into how the coordination takes place. It further examines the way in which these algorithms adapt to different environmental conditions such as communications connectivity, swarm size and rolerequirements. Chapters 7 and 8 turn to issues in swarm scalability both from the perspective of automated algorithm design and practical communications. In the former, it is shown that the transplantation of control policies generated through multi-agent reinforcement learning architectures is dependent on the way in which swarm agents observe their environment; to the best of the author’s knowledge, the first such result of its kind. This result allows the deployment of large swarms without the requirement to train all its members. In the latter, the breakdown of emergent behaviour is characterised as communications channel congestion increases, leading to a new metric for measuring emergence within such algorithms. As a complete work, the thesis makes several contributions to advancing the current state of autonomous swarm control through simulated and mathematical analysis. Where practicable, experiments were performed on real systems to further validate results in the real world. A desirable consequence of these contributions is an increase in trusted autonomy for systems that exploit swarm control.en
dc.description.dissertationThesis (Ph.D.) -- University of Adelaide, School of Computer Science, 2022en
dc.identifier.urihttps://hdl.handle.net/2440/135195
dc.language.isoenen
dc.provenanceThis electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at: http://www.adelaide.edu.au/legalsen
dc.subjectSwarmingen
dc.subjectSwarm intelligenceen
dc.subjectEmergent behaviouren
dc.subjectUAVen
dc.subjectUASen
dc.subjectMANETen
dc.subjectMADRLen
dc.titleSwarm Control for Mass UAS Developments in Disconnected Environmentsen
dc.typeThesisen

Files

Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
Fraser2022_PhD.pdf
Size:
10.99 MB
Format:
Adobe Portable Document Format
Download

Collections

Research Theses