Adaptation of the Random-Key Optimizer Framework for Agile Earth Observation Satellite Scheduling

This paper proposes an adaptation of the Random-Key Optimizer (RKO) framework to the Agile Earth Observation Satellite Scheduling Problem (AEOSSP), using instances generated by the EOSPython framework. The approach encodes solutions as random-key vectors and applies a greedy decoder that checks request, temporal, maneuver, and stereo-pair constraints. Experiments with one satellite indicate the feasibility of integrating RKO and EOSPython. Compared with the EOSPython DAG heuristic, RKO matched or improved the score in all nine evaluated instances, obtaining gains in seven of them, ranging from 0.61\% to 3.99\%. The results highlight the potential of RKO for handling complex AEOSSP instances and of EOSPython for generating reproducible evaluation scenarios.