Towards Autonomous Spacecraft: Developing a Multi-Application Autonomy Framework.

This thesis focuses on the development of an advanced autonomous decision-making infrastructure for ground-based satellite operations, designed to support mission planning, task scheduling and spacecraft control. The framework operates from Earth, providing both mission designers and operators with a comprehensive toolset to simulate, analyze, and optimize satellite operations. Currently, planning is largely manual, with limited support from automated tools, resulting in higher costs, especially for long-duration missions or large constellations. Implementing a fully autonomous planning framework has the potential to lower operational expenses while enhancing scientific and commercial outcomes. At its core, the system integrates a high-fidelity orbital simulator that models the satellite’s dynamics, orbital constraints, and visibility windows, enabling operators to evaluate scenarios and plan activities within feasible timeframes. Complementing the orbital simulator, the framework includes a CP-SAT-based task scheduler that models resource availability, operational constraints, and task priorities to generate optimized mission plans according to a custom objective function. This approach allows mission operators to anticipate conflicts, identify optimal plans, and adjust these plans dynamically to evolving mission objectives. The software architecture has been designed with extensibility and adaptability in mind, supporting application to both single-satellite and constellation scenarios, where the same approach is applied to multiple satellites simultaneously. This methodology maximizes mission return, particularly in Earth observation scenarios, as multi-satellite coordination increases and diversifies imaging opportunities. In all cases, validation of the generated plan and control of the spacecraft are intended to remain ground-based at this stage, to allow human verification before committing to the automatically planned satellite activities. As a natural extension, future work will focus on evolving the framework toward onboard autonomy, where decentralized decision-making would enable local and dynamic self-tasking of individual satellites. To this purpose, preliminary tests of the algorithm on a flight-equivalent data processor demonstrated compliance with memory and CPU usage constraints, marking a key step toward onboard deployment. To validate the proposed infrastructure, a representative case-study mission has been selected, with an initial implementation focused on orbital simulation and tasking optimization. Additional operational scenarios have also been considered to ensure the versatility and robustness of the system across multiple missions.