At NARC this week, our very own Bobby Baraldi will be giving an informal introduction to Kalman filtering, including a quick derivation and some implementations. To quote the infinitely wise Wikipedia, Kalman filtering is "an algorithm that uses a series of measurements observed over time, containing statistical noise and other inaccuracies, and produces estimates of unknown variables that tend to be more accurate than those based on a single measurement alone, by estimating a joint probability distribution over the variables for each timeframe." You can also check out Sasha's related paper (
https://arxiv.org/abs/1609.06369) for more information.