59. AI For Better Battery Design & Training Microrobots

DesignAI algorithm (inverse learning)

An AI algorithm analyzed electron microscope images of battery degradation to reverse-engineer the nanoscale physics, enabling researchers to design better lithium iron phosphate batteries with improved capacity and longevity.

“He used scanning transmission electron microscopes and advanced x-rays to take pictures of the actual batteries as they go through these charging cycles. And then use an AI algorithm to understand how they're degrading and response to different stimuli that it's seeing.”

Engineering & simulationartificial neural network (soft actor critic)less than 100,000 cycles

An AI controller learned to physically steer micro-robots in a real-world environment by receiving rewards for correct movements, eliminating the need for manual trial-and-error programming by engineers.

“they decided to switch that controller out from a human to an AI, an artificial neural network with a protocol they call a soft actor critic, which basically tells, is it doing what I want it to do or not? If yes, then I reward it, and if not, then you get a penalty.”