Automating Model Checking for Autonomous Systems

While autonomous systems offer great promise in terms of capability and flexibility, their reliability is particularly hard to assess. This paper describes research in the use of model checking to support the development of reliable autonomy software. In particular, it presents tools and techniques that we are developing to facilitate the integration of model checking into the main software development cycle. The basic approach is to translate highlevel models used by autonomy systems into the specification language of the SMV model checker, verify them using SMV, translate diagnostics back to the source language and visualize and explain those diagnostics. This approach has been applied to MPL models for the Livingstone fault diagnosis system and to TDL task descriptions for mobile robot systems.