Currently, the orbit estimator implemented in #295 has been tested very little. There are a couple key questions we need to answer and constants that need to be nailed down:
The GPS sensor values need to be updated to more realistic values (#142).
Once complete, we need to determine what order gravity model is the most efficient use of CPU time given the largest perturbations we expect to see which aren't accounted for in our model (most likely drag and or third body forces). This most likely means implementing a more advanced orbital dynamics model (see existing MATLAB implementation for an existing drag model).
With the proper order gravity model determined, process noise values during normal operations and around a thruster firing need to be settled on. The latter should be represented by incrementing the state covariance when delta v is applied to the orbital estimate.
Currently, the orbit estimator implemented in #295 has been tested very little. There are a couple key questions we need to answer and constants that need to be nailed down: