Actuator Fault Estimation in Robot Platoons

In controlled robot platoons, every subsystem needs information about the states of its neighbouring subsystems. On the other hand, actuator faults and communication errors have a considerable influence on the behaviour of the entire networked system. In this paper, an actuator fault diagnosis method is proposed that uses relative output information and it can be implemented without information interchange among the neighbouring subsystems. A subsystem model is proposed for the platoons, based on which the non-measurable states of the neighbouring subsystems can be determined by proper filtering techniques. In addition, the model is also used as the basis for designing a Proportional Integral (PI) observer for actuator fault estimation. Both the pole placement method and Linear Quadratic Estimator (LQE) are explored to determine the observer gain. The simulation results show that the model without network communication produces the same expected behaviour as the one that uses network communication. In addition, the proposed observer can correctly estimate the magnitude of the actuator fault.