Toward an optimal assignment of diagnosis method to mobile robots faults

Abstract

Fault Detection and Isolation (FDI) is a crucial task to ensure greater autonomy of mobile robots. This paper looks at the different techniques that serve the FDI process of mobile robots. In a first time, studied faults and characteristics that FDI methods must satisfy are listed. Then, 14 FDI methods are explained. They belong to 4 categories: Model-based, Knowledge-bases, Data-based and Material Redundancy approaches. Conditions of application, drawbacks and advantages are defined for each of them. A comparison between two methods (EKF and UKF) is made in simulation. The detection and isolation rate is good for GPS, IMU and Odometers faults but unacceptable for slip, blocked wheel and free wheel faults. This work is a first step toward designing a hybrid method to monitor the maximum of likely faults on mobile robots in real-time.