This study presents a reinforcement learning (RL) approach for reestablishing communication with deep-space satellites under unknown attitude determination and control system (ADCS) failures. When traditional fault-tolerant control methods cannot restore signal, the proposed RL controller acts as a last-resort measure by autonomously reorienting the satellite’s antenna toward Earth while charging the battery via solar panels. A generic reward function, designed for the RL-based method, enables the controller to adapt to diverse failure scenarios, including severe actuator noise, misalignment, and complete actuator failure. Simulations are conducted in the Basilisk environment and trained with the tonic framework and demonstrate ranging capabilities of the RL-based method across a wide range of ADCS failure modes. The controller not only learned two distinct tasks—antenna pointing and solar charging, but also learned how to point the antenna and maintain battery life in a single control scheme. The proposed method showed great success in noisy, misaligned, and underactuated cases and more notably, this approach obviates the need for prior knowledge of fault conditions or disturbance bounds. Future work will explore online RL algorithms for enhanced real-time adaptability and include hardware-in-the-loop simulations to further validate the method’s viability for deep-space missions.
","abstract_html":"<p>This study presents a reinforcement learning (RL) approach for reestablishing communication with deep-space satellites under unknown attitude determination and control system (ADCS) failures. When traditional fault-tolerant control methods cannot restore signal, the proposed RL controller acts as a last-resort measure by autonomously reorienting the satellite’s antenna toward Earth while charging the battery via solar panels. A generic reward function, designed for the RL-based method, enables the controller to adapt to diverse failure scenarios, including severe actuator noise, misalignment, and complete actuator failure. Simulations are conducted in the <em>Basilisk</em> environment and trained with the <em>tonic</em> framework and demonstrate ranging capabilities of the RL-based method across a wide range of ADCS failure modes. The controller not only learned two distinct tasks—antenna pointing and solar charging, but also learned how to point the antenna and maintain battery life in a single control scheme. The proposed method showed great success in noisy, misaligned, and underactuated cases and more notably, this approach obviates the need for prior knowledge of fault conditions or disturbance bounds. Future work will explore online RL algorithms for enhanced real-time adaptability and include hardware-in-the-loop simulations to further validate the method’s viability for deep-space missions.</p>","abstract_has_math":false,"creators":["Willoughby, Matthew"],"institution":null,"degree_name":"Master of Aerospace Engineering","degree_level":"Thesis - Open Access","degree_discipline":"Aerospace Engineering","degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-05-06T07:00:00Z","date_published":"2025-05-06T07:00:00Z","updated_at":"2026-07-27T19:26:16Z","subjects":["Satellite attitude control systems","Fault-tolerant control","Robust control","Adaptive control","Autonomous systems","Fault recovery","Safe‑mode control","Machine learning","Reinforcement learning","Deep deterministic policy gradient","Artificial Intelligence and Robotics","Astrodynamics","Navigation, Guidance, Control and Dynamics"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://commons.erau.edu/edt/901","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Willoughby, Matthew"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"thesis:degree_discipline","label":"Discipline","values":["Aerospace Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis - Open Access"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Aerospace Engineering"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Satellite attitude control systems","Fault-tolerant control","Robust control","Adaptive control","Autonomous systems","Fault recovery","Safe‑mode control","Machine learning","Reinforcement learning","Deep deterministic policy gradient","Artificial Intelligence and Robotics","Astrodynamics","Navigation, Guidance, Control and Dynamics"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://commons.erau.edu/edt/901"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This study presents a reinforcement learning (RL) approach for reestablishing communication with deep-space satellites under unknown attitude determination and control system (ADCS) failures. When traditional fault-tolerant control methods cannot restore signal, the proposed RL controller acts as a last-resort measure by autonomously reorienting the satellite’s antenna toward Earth while charging the battery via solar panels. A generic reward function, designed for the RL-based method, enables the controller to adapt to diverse failure scenarios, including severe actuator noise, misalignment, and complete actuator failure. Simulations are conducted in the Basilisk environment and trained with the tonic framework and demonstrate ranging capabilities of the RL-based method across a wide range of ADCS failure modes. The controller not only learned two distinct tasks—antenna pointing and solar charging, but also learned how to point the antenna and maintain battery life in a single control scheme. The proposed method showed great success in noisy, misaligned, and underactuated cases and more notably, this approach obviates the need for prior knowledge of fault conditions or disturbance bounds. Future work will explore online RL algorithms for enhanced real-time adaptability and include hardware-in-the-loop simulations to further validate the method’s viability for deep-space missions.
"]},{"key":"dc:title","label":"Title","values":["Satellite Reorientation Using Reinforcement Learning Under Unknown Attitude Failure"]}]}],"canonical_facts":{"dc:creator":["Willoughby, Matthew"],"dc:description.abstract":["This study presents a reinforcement learning (RL) approach for reestablishing communication with deep-space satellites under unknown attitude determination and control system (ADCS) failures. When traditional fault-tolerant control methods cannot restore signal, the proposed RL controller acts as a last-resort measure by autonomously reorienting the satellite’s antenna toward Earth while charging the battery via solar panels. A generic reward function, designed for the RL-based method, enables the controller to adapt to diverse failure scenarios, including severe actuator noise, misalignment, and complete actuator failure. Simulations are conducted in the Basilisk environment and trained with the tonic framework and demonstrate ranging capabilities of the RL-based method across a wide range of ADCS failure modes. The controller not only learned two distinct tasks—antenna pointing and solar charging, but also learned how to point the antenna and maintain battery life in a single control scheme. The proposed method showed great success in noisy, misaligned, and underactuated cases and more notably, this approach obviates the need for prior knowledge of fault conditions or disturbance bounds. Future work will explore online RL algorithms for enhanced real-time adaptability and include hardware-in-the-loop simulations to further validate the method’s viability for deep-space missions.
"],"dc:identifier":["https://commons.erau.edu/edt/901"],"dc:subject":["Satellite attitude control systems","Fault-tolerant control","Robust control","Adaptive control","Autonomous systems","Fault recovery","Safe‑mode control","Machine learning","Reinforcement learning","Deep deterministic policy gradient","Artificial Intelligence and Robotics","Astrodynamics","Navigation, Guidance, Control and Dynamics"],"dc:title":["Satellite Reorientation Using Reinforcement Learning Under Unknown Attitude Failure"],"thesis:degree_discipline":["Aerospace Engineering"],"thesis:degree_level":["Thesis - Open Access"],"thesis:degree_name":["Master of Aerospace Engineering"]},"updated_at":"2026-07-27T19:26:16Z"}