This paper proposes an approach for actuator fault detection & isolation (FDI) in a four mecanum wheeled mobile robots (4-MWMRs). The approach is based on a bank of unknown input observers (UIO) for linear parameter varying (LPV) systems. The FDI is challenging by considering faults with small amplitude and measurements with noise. Added to that, and considering the robot closed-loop control, the faults are compensated and they cannot be detected without a robust FDI algorithm. The objective is to detect and isolate actuator faults before the robot closed-loop deteriorates and leads to an unacceptable extent. keywords: 4-mecanum wheeled mobile robot, dynamic model, actuator faults. *Research leading to these results has received funding from the EU ECSEL Joint Undertaking under grant agreement n 737459 (project Pro-ductive4.0) and from the partners national funding authorities DGE.

In this paper, a combination of model-based and hardware redundancy methods is proposed for both sensor and actuator fault detection and isolation (FDI) of unicycle mobile robots. A focus is brought on robot drift-like faults on wheels and sensors. The goal is to detect and isolate the faulty component as early as possible. The proposed method is based on a combination of hardware redundancy and a bank of Extended Kalman Filters (EKF). Each filter is tuned for a specific fault, to generate residuals with different signatures under different component faults. The different signatures allow the fault isolation. Simulation results show that the proposed method allow to detect both wheels and sensors small drift-like faults and isolate them as early as possible.

No abstract

In this paper, the fault detection and isolation problem regarding actuation and sensing of a 4-mecanum wheeled mobile robot (4-MWMR) is studied. The challenge with respect to the current state of the art lies in detecting and distinguishing wheel sensor from wheel actuator additive faults for this kind of robots. An approach based on generating residuals is proposed. Sensor faults isolation is based on simply analyzing residual signatures which are different under each sensor fault. Due to omni-move properties, actuator faults are, however, more difficult to be isolated. More residual characteristics must be taken into consideration to achieve the isolation.