Earth-observation satellites require high-accuracy pointing to effectively collect data. This paper introduces a stochastic model predictive control framework that ensures robust pointing and coverage that explicitly accounts for both parametric uncertainty and persistent disturbances. Using offline sampling to approximate chance constraints, it enables real-time onboard implementation. High-fidelity simulations show that the approach reliably achieves pointing and coverage objectives, enhancing autonomy and mission reliability. Moreover, the results show that the stochastic scheme not only outperforms traditional predictive approaches that lack explicit guarantees, but also proves less conservative and computationally demanding than robust predictive methods.
Mammarella et al. (Thu,) studied this question.