Disturbance rejection in space applications: Problems and solutions

In the propulsive phase of planetary landing like Mars or Moon when only soft landing is the target, guidance, navigation and control are driven by an Inertial Measurement Unit and a radar altimeter/velocimeter. Their measurements are affected by bias and scale (factor) errors, which are aggravated by the vehicle axis to be tilted with respect to the local vertical for most of the descent trajectory, and by the attitude navigation error cumulated during the ballistic (and aerodynamic) flight after the orbiter separation. By complementing the Centre-of-Mass dynamics with appropriate disturbance state equations driven by noise vectors, and by estimating noise from the model error, scale errors and bias can be real-time retrieved as disturbance state variables, then eliminated from the tracking errors through disturbance rejection, under convergence conditions and sensor layout. The same result is proved cannot be achieved under pure feedback control. The result shows a well known fact: bias can only be eliminated by disposing of different sensors. A typical case study is the attitude control of drag-free satellites where fine accelerometers allow to reject wide-band drag torques at the price of the attitude drifting because of the accelerometer bias. Drift is cancelled by treating it a disturbance to be estimated by attitude sensors like star trackers. Simulated runs illustrate the results.