Multi agent control for space based interferometry

Abstract

Agent systems have been accepted and used advantageously by computer scientists since
\ntheir inception, but such systems have not been used so readily within the realms of control
\nengineering or robotics. The work contained within this thesis investigates separated
\nspacecraft interferometry in the context of a multi-agent system, under the influence of libration
\npoint orbital dynamics. The main focus is on the development of key agent skills,
\nincluding state estimation, guidance, control and decision methods to attain the desired
\nsystem output; within the consideration of decision methods, a comparison between centralized
\nand distributed decisions is made. Whilst mainly focussing on the development
\nof these skills, additional considerations pertinent to agent system development are also
\ndiscussed.
\n
\nA discrete time control method, integrating Kalman filtering with sliding mode control
\nand using potential function guidance to achieve velocity tracking with six degrees of
\nfreedom, is developed for the purposes of controlling agent motion. Whilst developed for
\nthe purposes of spacecraft agent control, the presented methods are equally valid to any
\nother vehicular agent system such as UAVs or AUVs if considering inter-agent regulation.
\n
\nCentralized and distributed decision methods are developed to enable appropriate autonomous
\nactions to be performed by the agent system. Primarily these actions include
\nselective attainment and regulation of a non-natural orbits relative to a central agent to
\nform an appropriate array configuration and instances of array reconfiguration to compensate
\nfor both failed agents and to maximize the mission duration.