Model-Based Programming of Intelligent Embedded Systems Through Offline Compilation

Many recent and future space missions point to the need for increased autonomy in spacecraft with an emphasis on more capable fault diagnostic systems. The most widely used fault diagnostic systems are rule-based. Rule-based systems have quick response to events and clearly present to engineers the predefined reactions to events. These systems, however, require engineers to manually generate all necessary rules and these do not convey the assumed model the engineers used to generate the rules. Contrarily, model-based systems eliminate the need to manually generate the rules. Programming complex embedded systems involves reasoning through intricate system interactions along lengthy paths between sensors, actuators, and control processors. This is a challenging, time-consuming, and error-prone process requiring significant interaction between engineers and software programmers. Furthermore, the resulting code generally lacks modularity and robustness in the presence of failure. Model-based programming addresses these limitations, allowing engineers to program reactive systems by specifying high-level control strategies and by assembling commonsense models of the system hardware and software.