Underwater structural damage inspection has mainly relied on diver-based visual\ninspection, and emerging technologies include the use of remotely operated vehicles (ROVs) for\nimproved efficiency. With the goal of performing an autonomous and robotic underwater\ninspection, a novel Tactile Imaging System for Underwater Inspection (TISUE) is designed,\nprototyped, and tested in this paper. The system has two major components, including the imaging\nsubsystem and the manipulation subsystem. The novelty lies in the imaging subsystem, which\nconsists of an elastomer-enabled contact-based optical sensor with specifically designed artificial\nlighting. The completed TISUE system, including optical imaging, data storage, display analytics,\nand a mechanical support subsystem, is further tested in a laboratory experiment. The experiment\ndemonstrates that high-resolution and high-quality images of structural surface damage can be\nobtained using tactile â??touch-and-senseâ?? imaging, even in a turbid water environment. A deep\nlearning-based damage detection framework is developed and trained. The detection results\ndemonstrate the similar detectability of five damage types in the obtained tactile images to images\nobtained from regular (land-based) structural inspection.
Loading....