Model-Based Autonomy for the Next Generation of Robotic Spacecraft

摘要

A novel approach to the design of reactive embedded software systems, called model-based autonomy, is generating significant interest in the space systems community. The goal of the model-based approach is to automate onboard sequence execution by tightly integrating goal-driven commanding with fault detection, diagnosis, and recovery capabilities. This paper describes Titan, a system-level autonomy framework that expands upon previous model-based approaches to configuration management, such as the Livingstone mode identification and reconfiguration system, which was demonstrated on the Deep Space One spacecraft. Titan is a model-based executive that estimates current spacecraft modes, detects and repairs failures, and executes commands, all within a fast sense-decide-act loop. This paper discusses the model-based software technologies implemented within the Titan executive. It describes a case study of the deployment of Titan to a representative space mission, including an overview of the various component models assembled, as well as the scenarios generated to verify proper execution of the software. Finally, it provides results from initial tests of the Titan implementation on these scenarios and models.