A decentralized fault detection and isolation strategy for networked robots

Abstract

This paper presents a distributed Fault Detection and Isolation (FDI) strategy, applied in conjunction with a distributed controller-observer schema, for a team of networked robots. Differently from other works in literature, the proposed FDI approach makes each robot of the team able to detect and isolate input faults of other robots even if not directly connected to it. The residual dynamics of the FDI observers are analytically investigated, and adaptive thresholds are derived to avoid the occurrence of false alarms in the presence of nonzero initial observer estimation errors. The approach is validated via numerical simulations in the case of time-varying centroid and formation control tasks.