An autonomous AI-enabled sensing system developed by Berkeley Lab researchers maps subtle magnetic, radiological, and optical signals from drones to locate “hot zones” rich in critical minerals such as neodymium. Credit: Berkeley Lab
AI and machine learning play a key role in characterizing and processing the critical minerals and rare earth elements needed for energy technologies, defense systems, and modern infrastructure. By understanding how minerals behave under different conditions, we can better locate them in the subsurface, improve ore processing, and more efficiently convert raw materials into usable forms.
EGD scientists are developing AI-enabled sensors, autonomous labs, and large language models to reveal the processes shaping mineral systems. Machine learning is being applied to predict the sequence of chemical steps and transformations a mineral or rare earth element undergoes, whether in nature or during processing, to account for thousands of possible pathways under varying conditions such as elevated temperature or reduced acidity. Autonomous laboratories are where these predictions are put to the test, allowing researchers to experiment and iterate at unprecedented speed and scale — bringing them closer to the specificity and selectivity essential to critical minerals exploration, given target minerals rarely occur in isolation but within complex geological settings alongside dozens of other elements.
In one example of this work, our researchers created a versatile tool deployable on terrestrial and aerial vehicles for surface waste mapping across diverse environments, reducing the time and cost of identifying “hot zones” of critical mineral concentrations. In another project, funded by Berkeley Lab’s Laboratory Directed Research & Development (LDRD) program, researchers are building an AI program — based on R&D 100 Award-winning software CrunchFlow — to help scientists develop chemical processes for extracting critical minerals from coal tailings, the fine-grained material left over from coal processing.
