Innovations in scientific instrumentation have pushed the boundaries of the observations possible by planetary science missions. These revolutionary tools allow scientists to capture data with higher spatial, temporal and spectral resolutions than ever before. Unfortunately, due to data-intensive nature of these instruments, mission planners are pressured to quickly analyze increasingly larger and more complex datasets to meet scheduled uplink timelines. This proposal introduces the development of machine learning powered tools that enable human-on-the-loop mission planning and data discovery. The goals of this project are to develop a framework that can rapidly and intelligently extract information from large planetary datasets, utilize that framework to create tools that can accelerate mission science planning, and implement that framework on existing data systems to improve their access, organization, and searching. This project is not looking to automate mission planning but rather create a critical tool used by mission planners for quick interpretation of data in order to minimize missed opportunities and efficiently plan operations. The project will utilize novelty detection machine learning systems to identify anomalous spectral and polarization features in high-dimensional planetary and astronomical data. In particular, the project will use traditional methods of anomaly detection, like Principle Component Analysis (PCA), as well as modern approaches, like deep learning for both ground-based an on-board applications. The development of this tool aligns with the NASA technology roadmaps goals Intelligent Data Understanding (11.4.2) and Advanced Mission Systems (11.4.5) as ways to handle the increase in data volume and density and enhance the capabilities of mission planners.
|Effective start/end date||8/1/20 → 7/31/24|
- NASA: Goddard Space Flight Center: $72,831.00
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.