Motion interference detection and identification in mobile robots using driving motor currents

A fault can occur any time during mobile robots operation and this may cause human injuries. To ensure the safety and the reliability of mobile robots, fault detection and identification (FDI) are necessary to be addressed. The faults in this study are defined as situations that the mobile robots collide with obstacles or the obstacle interferes mobile robot motion immediately. The objectives of this study are the fault detection of mobile robot collision without supervisory sensors. In contrary, we propose that the faults are detected by monitoring internal parameters of the service mobile robots. Moreover, we do not require the state equation. To classify the fault types, Pattern recognition Neural Network is applied. We evaluate the performance of proposed method through repeated experiments on a prototype. The finding of these experiments showed the effectiveness of Pattern Recognition Neural Network illustrated by the confusion plot. The proposed method can classify the fault types to target classes of the off-line simulated experiments.