自律型水中ロボットによる人工構造物の観測

Abstract

There is a great demand for Autonomous Underwater Vehicles (AUVs) not only for wide area surveys but also for condition surveys of artificial underwater structures such as breakwater caissons, pillars, ship hulls, and installations of offshore oilfields. This paper proposes a navigation method for AUVs operating around structures, whose major configuration is given in advance. By referring to sensory data and the map of the environment where the AUV is deployed, the method enables it to localize in real-time and follow pre-given way-points without any external help as needed for conventional acoustic positioning method. Probabilistic approach called “Particle filter” utilizes multi-sensor data to realize robust and stable navigation even at harsh condition where single sensor systems might fail. This method was implemented on the AUV ”Tri-Dog 1” and experiments were carried out around breakwater caissons at the mouth of Kamaishi bay, Iwate Prefecture. The vehicle took accurate images of the surface of a caisson, foot protection blocks, and rock mound. Mosaics of the bottom and the surface of the caisson are made from these images based on the precise vehicle's position estimated at each time of the mission and their quality verifies the performance of the method. This is the first time in world that caissons have ever been observed fully autonomously by an AUV.