Graduation Year

2023

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Mechanical Engineering

Major Professor

Tansel Yucelen, Ph.D.

Committee Member

Gordon Roesler, Ph.D.

Committee Member

Kyle Reed, Ph.D.

Committee Member

William Lee III, Ph.D.

Committee Member

Daniel Cunningham, Ph.D.

Keywords

Deterministic Optimized Decision Making, Finite-Time Control, Formation Control, Swarm Control, Model Reference Adaptive Control, Multiplex Information Networks

Abstract

Multiagent systems control designs offer numerous advantages to increasingly complex civilian, military, and scientific tasks ranging from intelligent satellite swarm controls or autonomous traffic controls to remote triage and repair or remote wildlife observation, where the performance of a single agent is insufficient to execute the task at hand. Traditional methods in multiagent systems controls, however, have been infeasible for use in scalable, large-group formation controls.

With this in mind, the contribution of this dissertation is to bridge the gap in knowledge through both theoretical and experimental validation of multiagent systems control architectures. It addresses the challenges presented by unreliable, brief, or compromised communications, sensing, and information exchange capabilities which can compromise the system’s decision-making capabilities. We provide a set of approaches to mitigate a number of common challenges in real world applications of multiagent systems controls.

In particular, a system consisting of a capable subset of leader agent(s) is given broader access to knowledge of the mission, while follower agents are given a limited access to commands, where the aim is to preserve system integrity through minimizing unnecessary interagent and out-of-system communications while meeting expected performance metrics. The approaches outlined above are validated through a series of theoretical proofs as well as simulations and hardware-in-the-loop experiments, demonstrating the first steps to future directions in multiagent systems control architectures in novel environments and various system uncertainties and constraints.

Share

COinS