Spaceflight under Uncertainty: Control, Optimization, and Learning
Apollo 11 Postdoctoral Fellowships at Purdue - Proposal
Research goals for the postdoctoral appointment
The proposed postdoctoral research will advance the theoretical and algorithmic foundations of spaceflight under uncertainty, addressing the growing need for safe, reliable, and autonomous spacecraft operations beyond Earth orbits. The postdoctoral fellow will develop new methods for space mission design and autonomy under operational uncertainties. Representative uncertainties of interest include, but are not limited to:
- navigational errors (e.g., spacecraft position and velocity),
- statistical sub-system anomalies (e.g., missed-thrust events, sensor failures),
- knowledge uncertainties (e.g., asteroid gravity fields, terrain detail for planetary/moon landing).
By unifying principles from astrodynamics, control theory, optimization, and machine learning, this research will produce novel understanding, rigorous theory, and scalable and verifiable algorithms for assured safety, reliability, and autonomy for space missions in challenging dynamical systems, (e.g., cislunar, interplanetary, and small-body environments). Such capabilities have the potential to enable ambitious space missions with unprecedented science return (e.g., cislunar autonomous transportation systems for sustained lunar science, searching life in the Jupiter/Saturn system, mapping and landing unvisited asteroids).
The postdoc researcher is expected to augment and complement research conducted in Prof. Oguri’s lab. The PI Oguri’s research involves astrodynamics, control theory, and optimization for risk-aware spacecraft autonomy, navigation, and trajectory optimization under uncertainty.
This project will complement it by incorporating one or more of the following expertise:
- scientific machine learning for quantifying uncertainty in unknown environments and/or characterizing the solution space and optimization landscape of mission design,
- reinforcement learning for enhanced robustness against model discrepancy between prediction and reality and/or adaptation to dynamically changing environments,
- data-driven methods for modeling nonlinear dynamics as a (quasi-)linear system (potentially in a lifted space) and/or creating surrogates of optimization results for global search,
- formal methods for verifiable and safe decision-making under uncertainty,
- diffusion/generative models for mission design and optimization from data,
- statistical inference for spacecraft perception to utilize unstructured and/or high-dimensional data inputs for autonomous navigation.
The project will expand the group’s research frontiers beyond astrodynamics, control theory, and optimization to scientific machine learning and statistical data-driven methods while integrating them with dynamical insights and rigorous control-theoretic guarantees. The project results will advance theory and algorithms for autonomous, uncertainty-aware spaceflight, contributing to NASA’s planetary science missions, DoD’s cislunar security objectives, and commercial endeavors in sustainable space operations.
Expected deliverables for the postdoctoral fellow
Through the project, the postdoctoral fellow is expected to:
- advance the fundamental understanding of synergies and interplay between astrodynamics, control theory, optimization, and machine learning,
- develop theoretical and algorithmic foundations for spaceflight under uncertainty that balance mathematical rigor and practical relevance,
- disseminate the discoveries and results to relevant academic/professional communities and public in the forms of conference presentations, journal articles, and relevant software implementations (to the extent reasonable).
In addition, the postdoctoral fellow will have opportunities to (and is expected to) interact with some of the PI’s collaborators across different sectors when appropriate, such as NASA, JPL, DoD, and industry partners.
Successful candidates possess a PhD in relevant fields (e.g., aerospace, electrical, or mechanical engineering, as well as applied math and computer science), with one or more of the complementing expertise listed above. Expertise in astrodynamics is a plus but not required.
Affiliated Faculty
Kenshiro Oguri, Assistant Professor
School of Aeronautics and Astronautics, Purdue University