Given that the system provides predictions of the state, we can use these to filter (e.g. completely remove), or downweight the impact of outliers. This suggests an iterative approach, whereby the observational mask gets updated if observations are very far away from predictions, or the uncertainty gets updated.
Given that the system provides predictions of the state, we can use these to filter (e.g. completely remove), or downweight the impact of outliers. This suggests an iterative approach, whereby the observational mask gets updated if observations are very far away from predictions, or the uncertainty gets updated.