Fender-Segmenting Densely-Connected AI Gives Drones a Shot at Keeping Port Facilities Safe
Designed to ease the maintenance burden on aging dock infrastructure, these smart drones can pick out fenders from their surroundings.
Researchers at Hanbat National University, Seoul's SISTECH, the Korea Institute of Civil Engineering and Building Technology, and Stradvision have made the first step in automating port maintenance — by giving drones a machine learning system for monitoring safety fenders on-the-wing.
"Periodic evaluation of port structures including outer walls and berthing structures with fenders is paramount to ensure their safety," the researchers explain of their work. "In accordance with the 2018 revision of the 'Special Act on Safety and Maintenance of Facilities' in Korea, fender damage and absence should be inspected and evaluated in the section of serviceability performance evaluation. However, most fenders are inaccessible on land and inspectors utilize floating boats for inspections."
Couple that with the increasing number of port facilities in Korea which are exceeding a 30-year service lifespan, and you've got a problem — but one that, the researchers claim, machine learning and drone technology can solve.
In their recently-published paper, the researchers showcase a system which combines uncrewed aerial vehicles (UAVs) with a novel computer vision system using a densely-connected encoder-decoder format inspired by what they describe as the "eccentric function" of the human eye. The drones, equipped with on-board cameras, took footage of the port facilities and used the machine learning system to identify fenders — segmenting them from their surroundings.
"We are planning to upgrade this model to the fender health inspection system," says Min Jiyoung, PhD, lead researcher on the project. "It will enable us to quantitatively detect damage such as missing sections or cracks from only UAV images. This UAV-AI combination technology will automatically evaluate the fender serviceability in the future, securing the safety of inspectors and reducing the time cost in the field."
The team's work has been published in the International Journal of Naval Architecture and Ocean Engineering under open-access terms.