A Deep Learning Method for Multiple Faults Detection and Classification of Unmanned Ground Vehicles

Abstract

Due to the increased complexity in actuators and sensors, unmanned ground vehicles have a better chance to generate faults in the course of operation. An untreated fault can result in a failure, which may lead to catastrophic consequences. In this paper, we propose a deep learning method using both input and output signals of the vehicles to learn the features of different faults reflected in the dynamic models of unmanned vehicles. We have applied the proposed method to detect and classify multiplicative and additive faults, as well as the faults that result in malfunction of the actuators. The results show that the proposed deep learning method can accurately detect and classify multiple types of faults, which are caused by different sources.