Agent control of cooperating satellites

Abstract

A novel, hybrid, agent architecture for (small)
swarms of satellites has been developed. The software architecture for each satellite comprises a
high-level rational agent linked to a low-level control system. The rational agent forms dynamic
goals, decides how to tackle them and passes the
actual implementation of these plans to the control layer. The rational agent also has access to a
MatLabmodel of the satellite dynamics, thus allowing it to carry out selective hypothetical reasoning
about potential options. This hybrid architecture
has been implemented on simulated Earth orbiting
satellites and hardware at Southampton’s satellite
hardware simulation lab. While the deployment of satellites in formation within an Earth orbit is clearly both interesting and useful, this paper takes the spacecraft swarm further a?eld. In particular, we here investigate the
rational agent autonomy problem, including collision avoidance, fault diagnosis and recovery, and cooperative behaviour of such spacecraft deployed to explore groups of Trojan Asteroids.

More information

Identifiers

Catalogue record

Export record