AUV Navigation Based on Multi-Sensor Fusion for Breakwater Observation

Autonomous Underwater Vehicles (AUVs) are suitable for underwater observation, since they can swim freely without an umbilical cable. This paper proposes a robust navigation method for AUVs operating around artificial structures such as piers, quays and breakwaters. This method consists of a State Estimator and a Guidance System, enabling localization in the configuration map as well as following pre-given waypoints without any support systems, except its own sensors. The state estimator is based on a particle filter where the vehicle’s state, horizontal position and heading, is stochastically updated in real time using multi-sensor data. This method was actually implemented to a testbed AUV “Tri-Dog 1” and a series of sea trials were carried out around breakwater caissons at Kamaishi bay in Japan. To verify the performance, mosaics of the breakwater are made from the observed images based on the estimated position and attitude of the AUV.