Automated reasoning on-board autonomous spacecraft

Traditionally spacecraft operations are automated by increasing the levels of automation in the ground segment. Certain missions, however, are only possible with high levels of on-board automation and autonomy with respect to the ground. These include missions to distant planets and bodies for which the light or radio signal travel time is too long to accommodate real time control and where the spacecraft spends long periods of time out of sight of ground stations. This paper reports on work done within the scope of the spacecraft autonomy research project which has as one objective to investigate architecture for on-board autonomy. The paper describes the autonomous spacecraft architecture and focuses on the on-board decision making mechanism. A highly decentralised and distributed architecture was adopted. The autonomous spacecraft is modelled as a collection of intelligent sub-systems that must co-operate to achieve goals. The decentralised nature of the system implies a heavy reliance on-communication for the co-ordination of actions, a mechanism for continued co ordination in the event of failure resulting in incomplete and uncertain spacecraft status information is described.