Integrating Low-Thrust Transfer Dynamics into POMDP Planning for On-Orbit Servicing

This research presents a Partially Observable Markov Decision Process (POMDP) framework for on-orbit servicing mission planning that explicitly models both high-thrust and low-thrust orbital transfers. The approach augments the state space with remaining propellant to account for mass-dependent dynamics, using precomputed matrices for high-thrust maneuvers and an analytical phasing model for low-thrust options. Policies are computed offline using the SARSOP algorithm and compared quantitatively to assess the operational feasibility and fuel efficiency of advanced electric propulsion systems in future servicing architectures.