DARPA wants to use AI to discover new rare minerals

Applications of AI

Safe access to rare earth minerals is a major national security issue, as the entire US economy is heavily dependent on them and most of the rare earth minerals discovered so far have been found in China. The Defense Advanced Research Projects Agency (DARPA) is partnering with a company called HyperSpectral, which applies artificial intelligence to spectroscopic data, which could be key to using satellites and drones to find minerals that are otherwise hard to detect.

Hyperspectral CEO Matt Terrue explained how it works in an exclusive interview. Defense OneSpectroscopy is the study of how matter interacts with light and other forms of radiation across a range of wavelengths. The solar radiation emitted by a particular mineral or material has unique meaning due to its unique molecular structure.

The company's focus so far has been on food safety: Want to know if your load of raw food contains deadly pathogens? Want to know about new outbreaks of drug-resistant strep? Spectroscopy can help spot bacteria that are invisible to the naked eye.

“The process currently used is [drug-resistant and drug-sensitive staphylococcus bacteria]”That's because you have to actually culture the bacteria and then apply antibiotics to see which antibiotics will kill the bacteria or not. Whereas now we test with a swab taken from the wound, and we usually get results in minutes instead of days.”

So where does AI come in? Terreul says, “Pure samples don't exist in nature. Nature is a very noisy place. So what we're doing with artificial intelligence when we build these models is looking for all the relationships that might be obscured by the noise.” [such as] That's when one part of the spectrum is confounded by other substances in it.”

And there are multiple types of spectroscopy that can't be easily combined into one data picture, but again, AI can help. Auditory data from human speech is very different from textual data about which combinations of letters and words most often appear together. But combining these can enable AI-driven transcription and translation. In theory, spectroscopy data from different sources could be just as useful.

“Absorption and reflectance [Fourier Transform Infrared Spectroscopy] “In Raman spectroscopy and surface-enhanced Raman spectroscopy, it's important to understand the spectroscopic response of those materials and be able to distinguish between different materials,” Terrell said.

What does it reveal? Terrell said the DEA has used similar technology and “was able to distinguish between cocaine coming from one cartel's territory in Colombia and cocaine coming from another cartel's territory.”

The cooperative agreement with DARPA is still in its early stages, and Terrell said the applications are numerous for the Defense Department to get a better understanding of where various materials are located. Spectroscopy, which can be performed by a small number of specific satellites, could aid in intelligence gathering, such as finding the presence of certain materials used in enemy or adversary equipment or vehicles.

In addition to improving access to rare earth elements, the Pentagon is also eager to move manufacturing of key weapons and supplies much closer to the front lines, rather than relying on supply lines in the Pacific that are extremely difficult to defend.

“The analysis of spectral data has a huge amount of applications and use cases — yes, a huge amount,” Teroul said.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *