Safety-Aware Optimal Attitude Pointing for Low-Thrust Satellites

In geostationary orbit, long eclipses and the seasonal variations in direction intensity of solar input can cause damage to sensitive equipment during attitude maneuvers, which may inadvertently point towards Sun. The requirement that transmitting receiving antennae remain pointed Earth creates further restrictions pointing directions. aim study is construct a novel geometric reinforcement-learning-based method determine guidance maneuvers maintain safe operational orientations throughout an maneuver. trajectory computed numerically using framing Pontryagin’s maximum principle applied vehicle kinematics global matrix Lie group representation on SO(3), angular velocities are shaped free parameters. values these parameters determined by reinforcement learning algorithm avoid forbidden areas while maintaining (modeled as subsets two-sphere all possible directions particular axis). model geosynchronous satellite demonstrated simulation.