Model-Based Reasoning for Complex Flight Systems
Marcello Balduccini, Michael Gelfond · 2005
This paper gives an overview of the design of a decision support system for the Space Shuttle that has the ability to find (usually in a matter of seconds) provably correct plans to achieve a given goal in the presence of single or multiple failures in the Reaction Control System (RCS). This tool includes a complete model of the RCS, including wiring and plumbing diagrams. Both the models and the reasoning modules were designed in the context of a project aimed at demonstrating the applicability of the knowledge representation declarative language A-Prolog, and of the answer set programming methodology in particular, to medium-size, knowledge-intensive applications. The project also demonstrated that A-Prolog allows a modular organization of knowledge, enabling knowledge module reuse, as well as testing (and debugging) of single knowledge modules, rather than of the entire knowledge base as a whole. The techniques used in the development of our decision support system are general enough to allow for the modeling of many other flight systems, and for the execution of reasoning tasks other than planning, including fault detection and diagnosis. I.