Selection-Induced Contraction of Innovation Statistics in Gated Kalman Filters
A new paper details how validation gating in Kalman filters can lead to a contraction of innovation statistics. The research shows that measurements below a certain threshold, when used for state updates, result in innovation statistics that are conditioned on the gate rather than nominal. This effect, which causes a deterministic contraction of the innovation covariance, is further exacerbated by nearest-neighbor association methods. AI