Fault Detection and Analysis of a Mobile Robot using Radial Basis Function Network

For autonomous mobile robot systems, which have uses in a variety of industries, including transportation and the military, accurate navigation is essential. Fault detection is vital since these mobile robots can have failures like wheel slides and accidents. This paper applies radial basis function networks (RBFN) to mobile robots in order to identify actuator failures. The neural network is trained and evaluated using MATLAB simulation to identify and classify faults in the unicycle differential drive (DD) mobile robot. Four different fault cases are examined in this study: simultaneous left and right wheel actuator faults, right wheel actuator fault, left wheel actuator fault, and the combination of no fault and simultaneous left and right actuator faults. For further investigation, the classification performance of the RBFN is also compared with that of a multi-layer neural network.