Formal Methods for Automated Diagnosis of Autosub 6000
This is a progress report on applying formal methods in the context of building an automated diagnosis and recovery system for Autosub 6000, an Autonomous Underwater Vehicle (AUV). The diagnosis task involves building abstract models of the control system of the AUV. The diagnosis engine is based on Livingstone 2, a model-based diagnoser originally built for aerospace applications. Large parts of the diagnosis model can be built without concrete knowledge about each mission, but actual mission scripts and configuration parameters that carry important information for diagnosis are changed for every mission. Thus we use formal methods for generating the mission control part of the diagnosis model automatically from the mission script and perform a number of invariant checks to validate the configuration. After the diagnosis model is augmented with the generated mission control component model, it needs to be validated using verification techniques.