Evaluation of anomaly and failure scenarios involving an exploration rover: a Bayesian network approach

Recent studies focused on the achievement of autonomy by spacecrafts, with the aim of avoiding the intervention of the ground control. In this sense, the ARPHA software prototype has been developed for the automatic failure detection, identification and recovery (FDIR), and is based on the on-board analysis of a Dynamic Bayesian Network (DBN) representing the system behaviour conditioned by the conditions of components and environment. In this paper, we describe the main functionalities of ARPHA, and we apply its FDIR capabilities to the power supply subsystem of an exploring rover, taking into account four scenarios leading to anomalies or failures. The DBN model of the system is described. Then, we test the execution of ARPHA, together with a rover simulator providing sensor data and plan data. In particular, we show the results of diagnosis, prognosis and recovery, returned by ARPHA when the scenarios occur.