Rubble to Rockets: Utilizing Scrap Metal and Machine Learning for On-Demand 3D Printing

Last week, we explored how the Defense Advanced Research Projects Agency (DARPA) is pushing the boundaries of additive manufacturing, and the momentum is not slowing down. On June 4th, the Worcester Polytechnic Institute (WPI) announced a $6.3 million grant from DARPA to develop a framework for turning scrap metal into 3D-printable material. The goal of the project, titled “Rubble to Rockets,” is to make it possible to produce critical components anywhere, even in resource-scarce environments like battlefields and search-and-rescue operations.

One key factor of “Rubble to Rockets” will be the use of machine learning. The WPI team will use artificial intelligence to identify materials like scrap metal and mixed alloys to understand how they react and bond together before being melted and mixed. 3D printing requires a precise understanding of materials, and typically, the process demands carefully controlled materials and repeated testing. However, that is not always possible when on the field or in an urgent situation.

VALIS team member evaluating metal (Credit: VALIS)

The researchers will use AI developed by a WPI PhD student to predict how a material will behave given various compositions. This will streamline the process for characterizing materials, allowing for faster production but without sacrificing durability or strength.