Distributed Fault Detection For UAV Formation Missions

A network of N uncertain aerial vehicle systems is connected via a Leader-Follower (L/F) configuration. These systems consist of control systems, sensors (proprioceptive/exteroceptive), and monitoring systems. If unanticipated faults occur in sensors or actuators then the objective is to design the monitoring systems such that the follower vehicles can timely detect faulty behaviour.This can be achieved by the inclusion of a Distributed Fault Detection and Isolation (DFDI) module. The proposed architecture involves the design of a bank of monitoring systems which generate residuals of the propriocetive and exteroceptive state that are compared against designed adaptive thresholds. Through such a procedure, it makes it possible to detect and infer on the faulty vehicle. The proposed methodology shows that the design of the architecture is decoupled from the size of the network, whilst decomposing the entire swarm into configurations of vehicles to be exploited. In particular, in the proposed design these configurations consist of two and three vehicles, respectively. The efficacy of the approach is shown through an example swarm scenario.