Date of Award
2026-05-01
Degree Name
Master of Science
Department
Computational Science
Advisor(s)
Tzu-Liang (Bill) Tseng
Abstract
The increasing density of satellites and space debris in Low Earth Orbit (LEO) has led to a rapid rise in conjunction events, creating significant challenges for collision avoidance and long-term space sustainability. Effective mitigation requires not only accurate prediction of collision risk but also reliable decision-making under uncertainty. Existing approaches often rely on deterministic thresholds or isolated optimization techniques, which are insufficient for addressing the stochastic and dynamic characteristics of orbital environments. This thesis presents a unified Forecast-Pareto-Conformal Markov Decision (FPC-MD) framework for risk-aware and uncertainty-informed collision avoidance. The proposed methodology integrates four key components: (i) data-driven forecasting of miss distance and collision probability using machine learning models, (ii) multi-objective Pareto optimization to generate feasible maneuver candidates considering both risk and operational cost, (iii) conformal prediction to provide statistically valid uncertainty bounds with guaranteed coverage, and (iv) a Markov decision layer to select optimal maneuver actions under uncertainty. The framework is evaluated using Conjunction Data Message (CDM)-based datasets, incorporating both observed and synthetically generated scenarios. Results demonstrate that the forecasting layer captures collision risk patterns effectively, while the conformal prediction layer achieves reliable coverage for both miss distance and collision probability. The Pareto optimization stage reduces the candidate maneuver space significantly, enabling efficient decision-making. The Markov decision layer selects optimal actions that balance risk reduction and maneuver cost, with the majority of events requiring no maneuver while critical cases are assigned corrective actions. Furthermore, a safe maneuver lane construction approach is introduced to characterize feasible operational regions under uncertainty, improving interpretability and safety assurance. The proposed FPC-MD framework provides a systematic and scalable solution for autonomous collision avoidance in space operations. By integrating predictive modeling, uncertainty quantification, and decision optimization, this work advances the development of intelligent space traffic management systems. The methodology is generalizable to other safety-critical domains involving decision-making under uncertainty and lays the foundation for future extensions, including reinforcement learning-based adaptive policies and multi-agent coordination.
Language
en
Provenance
Received from ProQuest
Copyright Date
2026-05
File Size
72 p.
File Format
application/pdf
Rights Holder
Mustafizur Rahman
Recommended Citation
Rahman, Mustafizur, "A Forecast-Pareto-Conformal Markov Decision Framework For Safe Maneuver Planning In Satellite Collision Avoidance Under Uncertainty" (2026). Open Access Theses & Dissertations. 4770.
https://scholarworks.utep.edu/open_etd/4770