Ensuring safety in adaptive flight controls systems is an ongoing challenge in aviation, especially as advancements in artificial intelligence and machine learning (AI/ML) trend upwards. Reinforcement learning is becoming more common in aerospace applications due to the ability to improve these models through training. While models such as reinforcement learning enable controllers to learn complex behaviors from interaction with the environment, their unpredictability in novel or disturbed conditions raises severe concerns in safety-critical domains. This research investigates the integration of runtime monitoring, a real-time assurance technique, with reinforcement learning-based flight controllers to ensure safety and reliability during flight. By supervising the system’s behavior during execution and enforcing formalized design constraints, runtime monitoring offers a vital middle ground between adaptability and control assurance. This thesis evaluates how runtime monitoring impacts safety compliance, mission success rate, and computation overhead in simulated flight scenarios consisting of waypoint tracking. By analyzing these results under both nominal and disturbed conditions, this work demonstrates the architectural tradeoffs between autonomous performance and runtime assurance, establishing a framework for the future of safe autonomous flight.
","abstract_html":"<p>Ensuring safety in adaptive flight controls systems is an ongoing challenge in aviation, especially as advancements in artificial intelligence and machine learning (AI/ML) trend upwards. Reinforcement learning is becoming more common in aerospace applications due to the ability to improve these models through training. While models such as reinforcement learning enable controllers to learn complex behaviors from interaction with the environment, their unpredictability in novel or disturbed conditions raises severe concerns in safety-critical domains. This research investigates the integration of runtime monitoring, a real-time assurance technique, with reinforcement learning-based flight controllers to ensure safety and reliability during flight. By supervising the system’s behavior during execution and enforcing formalized design constraints, runtime monitoring offers a vital middle ground between adaptability and control assurance. This thesis evaluates how runtime monitoring impacts safety compliance, mission success rate, and computation overhead in simulated flight scenarios consisting of waypoint tracking. By analyzing these results under both nominal and disturbed conditions, this work demonstrates the architectural tradeoffs between autonomous performance and runtime assurance, establishing a framework for the future of safe autonomous flight.</p>","abstract_has_math":false,"creators":["Zubyk, Andrew"],"institution":null,"degree_name":"Master of Software Engineering","degree_level":"Thesis - Open Access","degree_discipline":"Electrical Engineering and Computer Science","degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026-03-27T07:00:00Z","date_published":"2026-03-27T07:00:00Z","updated_at":"2026-07-27T19:26:22Z","subjects":["Flight Control System","Reinforcement Learning","Neural Network","Formal Methods","Runtime Monitoring","Navigation, Guidance, Control and Dynamics","Systems Engineering and Multidisciplinary Design Optimization"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://commons.erau.edu/edt/961","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Zubyk, Andrew"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical Engineering and Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis - Open Access"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Software Engineering"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Flight Control System","Reinforcement Learning","Neural Network","Formal Methods","Runtime Monitoring","Navigation, Guidance, Control and Dynamics","Systems Engineering and Multidisciplinary Design Optimization"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://commons.erau.edu/edt/961"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Ensuring safety in adaptive flight controls systems is an ongoing challenge in aviation, especially as advancements in artificial intelligence and machine learning (AI/ML) trend upwards. Reinforcement learning is becoming more common in aerospace applications due to the ability to improve these models through training. While models such as reinforcement learning enable controllers to learn complex behaviors from interaction with the environment, their unpredictability in novel or disturbed conditions raises severe concerns in safety-critical domains. This research investigates the integration of runtime monitoring, a real-time assurance technique, with reinforcement learning-based flight controllers to ensure safety and reliability during flight. By supervising the system’s behavior during execution and enforcing formalized design constraints, runtime monitoring offers a vital middle ground between adaptability and control assurance. This thesis evaluates how runtime monitoring impacts safety compliance, mission success rate, and computation overhead in simulated flight scenarios consisting of waypoint tracking. By analyzing these results under both nominal and disturbed conditions, this work demonstrates the architectural tradeoffs between autonomous performance and runtime assurance, establishing a framework for the future of safe autonomous flight.
"]},{"key":"dc:title","label":"Title","values":["Evaluating Runtime Monitoring for Reinforcement Learning-Based Flight Control"]}]}],"canonical_facts":{"dc:creator":["Zubyk, Andrew"],"dc:description.abstract":["Ensuring safety in adaptive flight controls systems is an ongoing challenge in aviation, especially as advancements in artificial intelligence and machine learning (AI/ML) trend upwards. Reinforcement learning is becoming more common in aerospace applications due to the ability to improve these models through training. While models such as reinforcement learning enable controllers to learn complex behaviors from interaction with the environment, their unpredictability in novel or disturbed conditions raises severe concerns in safety-critical domains. This research investigates the integration of runtime monitoring, a real-time assurance technique, with reinforcement learning-based flight controllers to ensure safety and reliability during flight. By supervising the system’s behavior during execution and enforcing formalized design constraints, runtime monitoring offers a vital middle ground between adaptability and control assurance. This thesis evaluates how runtime monitoring impacts safety compliance, mission success rate, and computation overhead in simulated flight scenarios consisting of waypoint tracking. By analyzing these results under both nominal and disturbed conditions, this work demonstrates the architectural tradeoffs between autonomous performance and runtime assurance, establishing a framework for the future of safe autonomous flight.
"],"dc:identifier":["https://commons.erau.edu/edt/961"],"dc:subject":["Flight Control System","Reinforcement Learning","Neural Network","Formal Methods","Runtime Monitoring","Navigation, Guidance, Control and Dynamics","Systems Engineering and Multidisciplinary Design Optimization"],"dc:title":["Evaluating Runtime Monitoring for Reinforcement Learning-Based Flight Control"],"thesis:degree_discipline":["Electrical Engineering and Computer Science"],"thesis:degree_level":["Thesis - Open Access"],"thesis:degree_name":["Master of Software Engineering"]},"updated_at":"2026-07-27T19:26:22Z"}