SAN DIEGO, Calif. (Jun. 19, 2019) Aboard the USS Midway Museum in San Diego, California, the Office of Naval Research Global (ONRG) TechSolutions held a demonstration of Topside Drone, a corrosion/anomaly detection sensor payload and processing scheme. U.S. Navy photo by Bobby Cummings
US Navy Turns to Drones, AI to Monitor Rust
The service has hired a Google partner to apply technology that has already changed cancer treatment and road repair.
The U.S. Navy has a $20-billion rust problem. Corrosion can hinder or even disable ships and aircraft at sea. But monitoring rust is expensive — $3 billion a year — and time-intensive for a Navy that’s already undermanned. So service leaders are turning to artificial intelligence software and drone photos to automate tasks and bring down the costs, Google announced today.
Simple Technology Solutions, a Google partner, will use “Google Cloud artificial intelligence (AI) and machine learning (ML) technologies on inspection drone-captured images to detect, prioritize, and predict its maintenance needs,” according to the statement.
Google’s ability to identify abnormalities in images with artificial intelligence is already helping to detect cancer, find problems in wind turbines, and spot potholes in Memphis streets. In essence, if you can build up a large enough database of images of a particular type, you can train the algorithm to spot things like cellular abnormalities or rust on the side of a ship.
“The only difference here is just the training inputs to the model, '' said Mike Daniels, the vice president for public sector at Google Cloud.
Daniels declined to comment on the amount of the contract.
STS will use drone photos and images in the public domain to build the model using Google Cloud AutoML, which lets organizations build custom AI products. Navy corrosion experts will help STS label and train the data using Google Cloud’s AI Platform Data Labeling Service. “This is one of those areas where it’s critical that the U.S. Navy can automate key processes like ship and vessel inspection,” Daniels said.
It’s also yet another sign of the warming relationship between Google and the Defense Department. The company helped DoD develop Project Maven, an AI program that helps human analysts sort troves of video footage to find targets. In April 2018, when Google's involvement in the program was revealed, many engineers and developers at the company protested. The company said it wouldn’t renew the contract. But it did draft an ethical principles list to lay out what sort of AI projects it would work on in the future.They’ve since won contracts to provide the Pentagon with cybersecurity services. Daniels declined to say whether the company was in discussion to offer similar object-identification for maintenance for other services or parts of the military.