Uncertainty Quantification
Uncertainty Quantification (UQ) is the science of how we quantify errors and confidence in our algorithms and numeric models. Our work focuses on UQ in the context of astronautics, and has overlap with our projects in Space Domain Awareness and Spacecraft Navigation. This work is multifaceted, and recent/ongoing projects include:
- Computationally efficient orbit state and uncertainty propagation
- Quantifying spacecraft risks (e.g., collision probabilities)
- Leveraging OUU in trajectory design
- Application to new and ongoing NASA and DoD missions
Some of the software developed through this research is used for intra-formation conjunction assessment for NASA’s Magnetospheric Multi-Scale (MMS) mission.
Current Projects
Uncertainty Propagation for Maneuvering Objects in Chaotic Systems
Sponsor: Air Force Office of Scientific Research
State-of-the-art Anthropogenic Space Object (ASO) tracking algorithms cannot robustly maintain custody of objects in cislunar space. One principal challenge is propagating uncertainty through multiple orbit regimes, each with different dynamic signatures and possible chaos. The goal of this project is to develop a computationally tractable and accurate method of uncertainty propagation for maneuvering objects that accounts for chaotic dynamics. In this collaboration with Prof. Ryan Russell, we will combine novel descriptions of a satellite’s possible maneuver profile and efficient orbit propagation routines to generate a predicted uncertainty for Bayesian multiple target tracking.
Non-Gaussian Estimation using Intrusive Polynomial Chaos
Sponsor: NASA through the NTGRO Program
Details coming soon!
Domain Splitting for Cislunar Orbit Uncertainty Propagation
Sponsor: Oden Institute for Computational Engineering and Sciences
Under a Moncrief Grand Challenge Award, this project uses methods of domain splitting to reduce the computational challenges of uncertainty propagation for trajectories in cislunar space. The domain splitting approaches use a combination of dynamical systems theory and sensitivity analysis approaches from the UQ community to customize the (sub-)domain for improved tractability. Through this approach, we are breaking down the global problem into a small set of easier-to-solve problems for uncertainty propagation.