How Putting A.I. Brains Into 3D Printers Will Change the Game for the Navy

One thing about airplanes—especially ones that fly from aircraft carriers, where they’re battered by saltwater and tough deck landings—is that they need lots of spare parts that are not always on hand. Instead of flying in new parts, though, future Navy ships may be able to make new ones to order.

Picutre an intelligent, laser-wielding robot that can analyze the damage and 3D-print the needed titanium alloy parts from an onboard supply of metallic dust. This is one glimpse of the future proposed by the Office of Naval Research (ONR), which today announced a two-year, $5.8 million contract to create a new generation of super-smart 3D printers. The printers would not only make parts on order wherever they are needed, but can “observe, learn and make decisions by themselves,” according to Lockheed.

The project is a collaborative effort between Lockheed Martin, Oak Ridge National Labs, Carnegie Mellon University, and four other partners. The team is starting with Ti-6AI-4V, a common titanium alloy used in aerospace. If the project succeeds, it could demonstrate how artificial intelligence could change everything you know about manufacturing.

It’s easy to see how manufacturing new parts on the spot could change the game for the U.S. Navy. But there’s a dirty little secret about 3D printing that limits its use with machines that endure extreme stress, like spacecraft and airplanes.

Consider the materials themselves. Aerospace-grade metals, including several recipes of titanium alloy, are supplied by foundries and have well-known characteristics. This raw metal comes with guaranteed strength, porosity, and thermal tolerance characteristics. Not so with 3D printed metal, which is made layer-by-layer on the spot.

What engineers call the microstructure of the metal, meaning the size, shape, and orientation of the grains, for example, is not guaranteed from a 3D-printed metal part. That piece could look identical to a traditionally manufactured one but perform differently.

“With traditional, subtractive manufacturing you have the same properties in the final part,” says Glynn Adams, a 25-year welding veteran who works for Lockheed in Michaud, LA . “But with additive manufacturing the material and mechanical properties are not as well understood.”

ONR has a plan, though. By outfitting the 3D-printing robot arms with commercial sensors, the lab is hoping to create a database that ties 3D printing processes and conditions with the resulting microstructure. The data will create predictive models that will enable 3D printing machines to create parts with foundry consistency, but from anywhere. “We have to build quality into the part,” Griffith says.

This is where A.I. comes into play. Machine learning algorithms allow these 3D printers make adjustments on their own to match the material qualities the military is looking for. It’s manufacturing by wire: Simply provide the shape and needed performance properties of the metal, and the 3D printer will take it from there. In other words, the printers will train themselves to make decisions on how to build things.

“IT COULD ENABLE ON-ORBIT MANUFACTURING"

“When you can trust a robotic system to make a quality part, that opens the door to who can build usable parts and where you build them,” says Zach Loftus, Lockheed Martin Fellow for additive manufacturing.

A fleet of future Navy ships or spacecraft could learn from each other’s experience by feeding the data from each robot back to a central brain. “The project with ONR is at the inception of this,” Griffith says. After all, he works within Lockheed’s space systems group, opening up another vista for 3D-printed parts. Intelligent 3D printers open up new ways of building things in space, saving money on launch costs and nearly impossible quality control.

“It could enable on-orbit manufacturing,” Griffith says. “Think about the freedom additive manufacturing might enable when you can trust the material properties.”