Brain-Like Chip Tracks Flying-Taxi Batteries Using Almost No Power

Electric air taxis — small aircraft that take off and land straight up, no runway needed — live or die by knowing exactly how much battery power they have left. Today that job is usually done with the same kind of AI that powers chatbots: powerful, but hungry for computing power and electricity. A new paper describes a different approach: a chip that estimates battery charge the way a brain does, by firing simple on-off pulses instead of crunching big piles of numbers.

The trick is what the chip doesn't do. Standard AI battery estimators perform about 67,700 multiply-and-add calculations every time they update. This design needs roughly 3,500 simple additions and zero multiplications. That's like swapping a calculator for counting on your fingers — except it still gets close to the right answer. The whole system was built onto a cheap, automotive-grade chip you can buy off the shelf, and it ran a real 491-step flight segment exactly as predicted.

There is a real trade-off. The new method is about 2.45% off on battery-level estimates, versus 1.74% for the best standard AI. For a flying vehicle, that gap matters. But the pulse-based chip degrades 1.6 times more slowly when sensors get noisy, which is exactly the kind of messy condition real flights create. And it sips power: about 1.3 microjoules per reading, averaging 0.65 microwatts — far less than a digital wristwatch.

The catch: this is a single research result on one public dataset, not a flight-certified product. Aircraft safety rules are strict, and a 2.5% battery error still needs proving across many conditions. Still, it shows a path toward lighter, cooler battery brains — useful not just for air taxis but potentially for electric cars, drones, and anything battery-powered that can't afford to waste energy on thinking.