Fault Diagnosis for Industrial Robots

In the last decade considerable research efforts have been spent to seek for systematic approaches to Fault Diagnosis (FD) in dynamical systems. Special attention has been put in robotic systems, especially for those operating in remote or hazardous environments, where a high degree of safety as well as self-diagnostics capabilities are required. On the other hand, the development of effective strategies of fault diagnosis for robot manipulators operating in an industrial context is a critical research task. Several FD techniques for robot manipulators have been proposed in the literature, although the problem of their application to industrial robots has not been extensively investigated. In this chapter different discretetime observer-based approaches to FD for mechanical manipulators are presented and critically compared. First, a rough FD technique is considered, which is based solely on the prediction capabilities of the manipulator dynamic model. Next, an observer-based technique is presented, where a robust time-delayed compensation is introduced to cope with disturbances and modeling uncertainties. Finally, two different observer-based schemes are developed, where the uncertain terms in the model are dynamically estimated: the first scheme is based on the recursive estimation of the uncertain terms, while the second one adaptively estimates the parameters of a suitable parametric model of the uncertainties. All the considered schemes are experimentally tested on a six-degree-of-freedom industrial robot and the performance are critically compared each other.