Fault Tree Analysis of Sensor Technologies for Autonomous UUV Navigation

Autonomous unmanned underwater vehicles (UUVs) are increasingly used for inspection and cleaning tasks. While automating these tasks could greatly reduce the cost, it requires reliable feedback from position and surroundings. Both internal effects and different physical properties affect sensors, resulting in inaccurate feedback if not handled correctly by the navigation system. In this study, an overview of these effects and properties are examined for the most common sensor technologies used for underwater navigation. A fault tree analysis (FTA) is conducted to get knowledge about how the sensor faults, as a result of these effects, affect automated near-structure and off-structure missions, respectively. Moreover, experiments are carried out with a high-resolution sonar and stereo camera to compare the measurement accuracy at different distances. The sensor comparing test shows that cameras can, in some cases, be insufficient to use as the only sensor for obstacle avoidance. It is concluded that the sensor criticality is case-specific; in general, especially faults on attitude feedback are severe for an acceptably-working navigation system and should therefore have high priority when selecting the robotic sensors.