Ah, so you’re missing the STMs when the measurements are rejected? Remember from our earlier discussion, the EKF processes a measurement at a time, regardless of whether the measurement is rejected … if the measurement is rejected then the updated state is just the predicted state. You need to extract the STM at all time-steps, but it looks like you’re getting nulls when the measurement was rejected? I’ll take a closer look at the orekit/hipparchus internals to see if that’s expected behaviour … we might need to make some changes.
1 Like