Optimal Planning and Scheduling for Satellite Servicing Under Partial Observability

This study proposes a novel sequential modeling and optimization framework for event-based On-Orbit Servicing (OOS) missions under partial observability and stochastic servicing demands. The problem is formulated as a Partially Observable Markov Decision Process (POMDP) in which a servicing spacecraft performs optimal maintenance decisions based on the uncertain health state of target satellites. The main contribution is the development of a novel event-based modeling framework for maintenance-oriented OOS missions, that explicitly incorporates partial observability and imperfect monitoring of satellite health state. This framework captures dynamic orbital travel times and variable spacecraft service durations, enabling more robust and adaptive mission planning in the presence of both epistemic and aleatory uncertainties. The Successive Approximations of the Reachable Space under Optimal Policies (SARSOP) algorithm is adapted to efficiently solve the proposed POMDP event-based planning model. The framework is validated through case studies involving satellites in Medium Earth and Geostationary Orbits. The framework helps mission planning by contrasting the benefits of acquiring additional information about the satellite's health state against executing optimal servicing actions based on current knowledge.