Skip to main content
GE Global Research

Brothers In Arms: These Robots Put A New Twist On 3D Printing

Todd Alhart
December 15, 2017
In October, engineers at Avio Aero, a GE Aviation company, used a futuristic process called cold spray to repair a gearbox on the GE90, the largest and most powerful jet in the world. The technology uses a special supersonic nozzle attached to a robotic arm. The nozzle shoots a barrage of tiny metal specks at four times the speed of sound at metal components like the gearbox. The bits land with such force that the solid-state particles start behaving like liquid, form a new layer and restore worn-out sections without changing the mechanical properties of the original.
The feat was a major milestone in the world of additive manufacturing, a family of new technologies that includes 3D printing. But GE engineers barely stopped to take a victory lap.

Leo Ajdelsztajn and his team of scientists at GE Global Research are already taking the technology to the next level. They are working on ways to use cold spray to build new parts instead of just fixing them. “One of the advantages of cold spray as an additive manufacturing modality is that we are not confined to a specific build volume or size,” says Ajdelsztajn.



Additive manufacturing usually happens in a confined space. GE recently introduced a beta version of the world’s largest 3D printer for metals, which can build parts as large as 1 meter along each of three axes.

Cold spray could work on a larger scale. Recently, Ajdelsztajn and his team added a second robotic arm and machine learning into the mix. Moving in perfect sync, one arm holds the part, always moving it to a precise location, while the other sprays it with powdered metal, adding material to it. The robots move together in a fully coordinated 12-degrees-of-freedom space, which means they can move not only forward and backwards and up and down, but they also can tilt and pitch in different directions. The team already used the experimental design to build an airfoil for a jet engine.

For this technique to be effective, and repeatable on a large scale, the robots have to move with exact precision. If one is off by as little as the width of a human hair, Ajdelsztajn says, the entire part could be ruined.

That’s why the team reached out across the hall to Joe Vinciquerra, a GE scientist exploring ways to include artificial intelligence and machine learning into 3D printing and other manufacturing technologies. The idea is to make the robots learn as they work and improve with every new part they make. “Imagine painting the same picture 40,000 times per year,” Vinciquerra says, referring to typical production numbers of additive parts. “Not every picture will be identically the same — even if a machine is doing it. Some will be better than others, and we can learn from those minute differences. By applying those changes in real time, the quality of every new painting increases.”

Vinciquerra says that the robots should be able to improve as time goes on and limit mistakes by analyzing the set of instructions the robots followed each time they made a part.

There are many different modalities of 3D printing, some using lasers to build a part, for example, and others using an electron beam. Adelsztajn says that cold spray is “a different brush in the painter’s kit. One way in which we are building our additive toolbox is by looking at how each additive technique balances the others — an artist wouldn’t limit themselves to one color of paint and one size brush.”