Verification of Autonomous Systems for Space Applications
Autonomous software, especially if it is based on model, can play an important role in future space applications. For example, it can help streamline ground operations, or, assist in autonomous rendezvous and docking operations, or even, help recover from problems (e.g., planners can be used to explore the space of recovery actions for a power subsystem and implement a solution without (or with minimal) human intervention). In general, the exploration capabilities of model-based systems give them great flexibility. Unfortunately, it also makes them unpredictable to our human eyes, both in terms of their execution and their verification. The traditional verification techniques are inadequate for these systems since they are mostly based on testing, which implies a very limited exploration of their behavioral space. In our work, we explore how advanced V&V techniques, such as static analysis, model checking, and compositional verification, can be used to gain trust in model-based systems. We also describe how synthesis can be used in the context of system reconfiguration and in the context of verification.