The number that tells you a battery is at 47% is, strictly, a guess. Unlike a fuel tank, a cell has no float you can read. The state of charge has to be inferred from things you can measure — voltage, current, temperature — and inference is where battery management systems earn their keep.
There are two basic methods, and every BMS blends them. The first is coulomb counting: measure the current flowing in and out, integrate it over time, and track how much charge you've moved. It's accurate short-term but drifts, because small measurement errors accumulate. The second is voltage-based: a cell's open-circuit voltage maps to its charge level, but that map shifts with temperature, age, and load, so it's unreliable in the moment. Combine them and each covers the other's weakness.
The patents are claims on how to do that combination well. Mitsubishi Electric's US10254345B2 describes a state-of-charge estimation apparatus and method. Atieva — the company behind Lucid — holds US10690725B2 on battery state-of-charge estimation. Silicon Works' US10295604B2 claims an estimation method tied to the BMS itself. The recurring CPC code across them, G01R 31/367, is the patent office's label for estimating battery condition.
One analogy and I'll drop it: estimating state of charge is dead reckoning at sea. Coulomb counting is tracking your speed and heading to plot where you should be; the voltage model is the occasional star sighting that corrects your accumulated drift. Neither alone gets you home; the navigator who blends them does.
Why this matters beyond engineering: state-of-charge accuracy is what lets a carmaker safely use more of a battery's real capacity. A BMS that's uncertain has to leave a buffer — reserve capacity it never touches — to avoid over-discharge. A better estimator shrinks that buffer, which means more usable range from the same physical cells. The estimation algorithm is, in effect, free energy density.
These are method patents, not guarantees of accuracy in every condition, and SOC estimation remains genuinely hard at the extremes — near-empty, near-full, and very cold. But the cluster shows the value clearly: the cheapest way to get more out of a battery is often not new chemistry but a smarter estimate of what's already in it.