Data Abstraction Architecture for Spacecraft Autonomy

Spacecrafts generate huge amounts of data. A significant challenge for autonomous control systems and human operators is ensuring that the right data (and combinations of data) are available at the right time for control and decision-making and ensuring that the data is at the right abstraction level. In this paper, we describe a data abstraction architecture that provides a canonical way to assemble and interact with data abstractions. We have developed an open, flexible toolkit that allows end users to build data abstraction networks. Two use cases were successfully tested against a Lunar habitat simulation, demonstrating that high level state information can be generated by the data abstraction architecture and used by a high level controller. Our approach improves the process of both control and monitoring of space systems by separating controls and displays from data abstraction.