A model based self-diagnosis system for autonomous underwater vehicles using artificial neural networks

Summary form only given. In the underwater environment, installation of a proper scheme to cope with unexpected troubles is essential for autonomous underwater vehicles (AUV) to carry out their mission out of human\´s reach. This paper proposes a model based approach to self-diagnosis for AUV in order to supervise whether the vehicle operates itself in an appropriate way. The proposed self-diagnosis is carried out based on a dynamics model of an AUV and an active mechanism to get desirable information for diagnosis. The dynamics model is constructed by an artificial neural network taking advantage of its flexible learning capability. When a sensor is found to be defective, dead reckoning using its corresponding output of the dynamics model can be introduced in an attempt to cope with the defect. The performance of the proposed system was examined by implementing it to "The Twin-Burger", an actual test-bed AUV. It is shown that the system detects failures of onboard sensors and actuators without introducing extra sensors for the detection, and then selects a proper action scheme to minimize the damage to the AUV.