#HTE

Engineers Develop a Hyper-Compressible Material Using Artificial Intelligence

You may not realize that among more conventional applications of artificial intelligence like apps and search engines, emerging technologies are transforming yet another unexpected area of design: materials. A fascinating new material study released by Delft University is showing how machine learning may upend our assumptions of how materials are capable of behaving.

The Delft study, led by assistant professor of materials science and engineering Miguel Bessa, has developed a new meta material that transforms brittle polymer materials into ultra compressible forms—to understand what this means for the future of product design, Bessa says with an innovation like this, “everyday objects such as bicycles, dinner tables and umbrellas could be folded into your pocket.”

While it’s difficult to imagine fitting an entire bicycle in your back pocket, the material has been developed in scales ranging from macro to nano and shows great promise.

So how do researchers utilize artificial intelligence to develop new materials? As Bessa describes it, “Traditionally, you would go into the lab and by trial and error you would try to find a material that could do this… What Artificial Intelligence allows you to do is to revert that process. Instead, you do simulations in a computer and you let AI learn how the material is behaving and what is changing. It basically gives you a treasure map. And us, the scientists, just need to find the treasure after we find the map.”

Bessa discusses in the study how perhaps the most important discovery within this investigation is how technologies like machine learning can create vast new opportunities in the design space: “Machine learning creates an opportunity to invert the design process by shifting experimentally guided investigations to computationally data-driven ones…data-driven science will revolutionize the way we reach new discoveries.”

As for this new metamaterial, research is still being conducted to refine previous iterations of the material configurations. For those who are computationally literate, the researchers have even shared the code used to in the study as an open source so that people outside of Delft University can utilize and improve it. Let’s hope all of these efforts in combination can inch us that much closer to our “bike-in-a-pocket” dreams.