Innovative Software for Autonomous Fault Detection and Diagnosis on Space Systems

In the last fifteen years the space community has increased its involvement and its efforts in the so-called onboard autonomy. The most relevant application of the studied methodologies and algorithms, mainly taken from the very wide Artificial Intelligent research field, has been the Remote Agent Experiment [1][2]. A lot of definitions of autonomy exists but in general it is possible to recognize that an autonomous spacecraft is capable of managing all the operations to be performed in order to accomplish all the mission objectives, without any intervention by human operators. It means to be able to decide what to do, to construct a plan for reaching the mission objectives consistent with the constraints and to really execute it. During the execution of the plan, some unexpected events may occurs, so the spacecraft must rapidly react to them; in particular when some malfunctions happens, the system should be able to make a correct diagnosis identifying the causes of the fault, and consequently to take some recovery actions; this last issue is inserted into the fault detection, identification and recovery (FDIR) research field. There are many motivations for increasing the autonomy level on-board a space system: the most relevant are the reduction of mission overall costs, the increasing of its scientific return, and the increment of the degree of reliability of the whole mission. In particular, FDIR methodologies are intended to provide autonomous agents with fault-tolerance and robustness of execution in an unpredictable environment. More in details, FDIR technology has the following goals: • monitoring the system behaviour in order to detect some malfunctions analysing the deviation of some sensed variables from a nominal or forecasted behaviour; • reasoning on the causes that could have brought to these symptoms, identifying the set of possible failures; • when the fault is isolated, the system must establish a certain strategy to re-establish its correct functionality, according to mission goals and constraints and also according to the actual configuration.