Application of Fault Detection and Isolation Techniques on an Unmanned Surface Vehicle (USV)

The detection and the isolation of a common fault occurred in an Unmanned Surface Vehicle (USV) is presented. A data-driven, model-free technique based on the Principal Components Analysis (PCA) technique is used to formulate the fault detection problem. This choice is particularly suited for applications on underwater robotic vehicles where, in general, dynamic models are not available or not appropriate for fault detection purposes. Tests performed on telemetry data acquired during field operations show that the presented approach is practical and effective to cope with unexpected environmental situations.