←back to thread

1070 points dondraper36 | 1 comments | | HN request time: 0.286s | source
1. moffkalast ◴[] No.45072490[source]
Nooo you can't use deterministic math, you gotta use kalman filters and particle filters and factor graphs and spend three months tweaking parameters to get a 5% improvement! /s

Like alright in some situations that's the only thing that could possibly work, but shoving that complexity into every state estimator without even having a way to figure out the actual covariance of the input data is a bit much. Something that behaves in an expected way and works reliably with less accuracy beats a very complex system that occasionally breaks completely imo.