Synthesizing Diagnostic Explanations from Monitoring Data in Multi-Robot Systems

The paper discusses the issues concerning monitoring and diagnosis of a multi-robot system involving a team of mobile robots providing services in an environment which is partially observable via a net of sensors. The paper describes the techniques for monitoring on-line the progress of the actions performed by robots and detecting deviations from the nominal actions behaviour. Because of the partial observability of the system, many possible evolutions are tracked by the monitoring and these pieces of information are synthesized by a diagnostic interpretation module in terms of robots faults and troublesome interactions among robots. The results of the diagnostic interpretation are displayed to a human supervisor via a graphical interface. The paper reports experimental results concerning the performance and the competence of the Diagnostic Module gathered by exploiting a simulator of the specific RoboCare project.