BV concludes proof of concept test for AI-led corrosion detection

Bureau Veritas successfully tested the use of the AI solution in real survey conditions. The test was conducted during a hull inspection of a water ballast tank on a bulk carrier in partnership with the RIT operator, MaDfly Marine Drone Services Ltd, in Dunkirk, France.

During the hull inspection, BV confirmed the value of the solution to support the decision of the surveyor during his general visual inspection and close-up survey of the hull in a number of important ways:

• Real-time information: capacity of the application to provide results on the video in real-time.
• Offline: capacity of the application to run offline without any connection to internet.
• Real survey conditions: capacity of the application to be deployed on the field with the computing capacity of a normal computer.

Corrosion represents one of the largest through-life cost components of ships. Ship owners and operators recognize that combating corrosion significantly impacts vessel reliability, availability, and through-life costs. Through periodic surveys, classification surveyors assess the severity of the corrosion on hull structure and monitor its evolution during the entire lifecycle of the asset to decide when some steel needs to be replaced in order to maintain the structural integrity of the hull.

This proof of concept is a first step for Bureau Veritas to enable the ship owners and ship managers to anticipate the necessary repairs. Moving forward, BV will further improve the solution and include new functions, leveraging its surveyors’ experience and data collected from digital technologies such as remote inspection techniques.

With a more holistic perspective, this development is a continuation of BV’s Digital Classification program that aims to transform its classification operating model through 3D-classification, remote and augmented surveys, and predicted and optimized survey schemes. Bureau Veritas invests in the development of innovative AI solutions, based on deep learning, for corrosion detection, and severity assessment.