ADCS.estimators.process_model module

Pure estimator-facing state propagation.

This module is the narrow boundary around the satellite’s historical integration API. Filters call propagate_state(); they do not need to know how noiseless_rk4 represents physical versus augmented estimator state.

ADCS.estimators.process_model.propagate_state(state, satellite, control, dt, orbital_state_start, orbital_state_end, *, midpoint_orbital_state=None, quaternion_integrator='rk4')[source]

Return deterministic propagation without mutating state.

The physical state is propagated by satellite.noiseless_rk4, whose default error mode applies the actuator biases currently held by the actuator objects (an estimator synchronizes its estimated biases there via EstimatedSatellite.match_estimate) and adds no noise. For an EstimatorState, wheel momentum is propagated while estimated bias and disturbance blocks, covariance, and process noise are copied unchanged. Those nominal parameter blocks are constant under this deterministic model; their uncertainty evolves separately through the process-noise model.

quaternion_integrator is "rk4" for normalized component RK4 or "cg5" for the satellite’s Lie-group variant.

Parameters:

  • state (State)

  • satellite (Any)

  • control (Any)

  • dt (float)

  • orbital_state_start (Any)

  • orbital_state_end (Any)

  • midpoint_orbital_state (Any | None)

  • quaternion_integrator (str)

Return type:

State