In this paper, a method for autonomous visual inspection of intricate underwater structures by AUVs (Autonomous Underwater Vehicles) is proposed. AUVs can perform the inspection task by dead reckoning navigation and error correction by using a sacrificial anode as a navigational aid. The parallel lighting method is proposed to identify an inspection target that may be covered with marine organisms while using a limited amount of computing power. Simulations and field experiments have been carried out to demonstrate the efficiency of the proposed technique.
With regard to the underwater environment, most of underwater structures are huge and they too require automatic inspection. Especially, extensive visual inspection of underwater structures such as the bottom of Mega-Float (National Maritime Research Institute 2001), dams, concrete caissons at ports and offshore structures having complicated surfaces is known to be very difficult to accomplish by manual inspection performed by ROVs (Remotely Operated Vehicles) or divers (Japan Society of Naval Architecture 1999; Japan Acoustic Committee 1999; T. Ura et al. 1994; David Sisman 1982). For example, the bottom shell area of Kamigoto's oil storage tanker (H. Sugimoto 2002) is more than 200,000m2. Because of its huge area, full visual inspection of the bottom shell is almost impossible by manual inspection. Moreover, its huge area increases the length of a diver's emergency escape route from the center location and also puts a heavy load on the ROV's tether handling system. When ROVs are used to inspect its bottom shell, they may often lose their way unless they are provided with comprehensive support such as an LBL acoustic positioning system. In this paper, an autonomous visual inspection method for AUVs is proposed. Free movement allows the AUVs to investigate wide and complicated structures with a minimal support system on the deck.