Spaceborne Hyperspectral RSO Imaging Under Downlink Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hyperspectral imaging for space situational awareness faces challenges due to data downlink bottlenecks created by the large volume of hypercube data produced by HSI sensors, necessitating improved systems and methods for RSO characterization.
Innovation Solution
A combined onboard and ground-based machine learning approach for hyperspectral data processing, involving onboard data reduction and compression, followed by ground-based enrichment and optimization, using sensor fusion and advanced algorithms to enhance RSO and component spectra identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hyperspectral imaging sensors are used to collect data of resident space objects, then identification accuracy is improved, but data volume increases causing downlink bottlenecks
Solution Approach 1:
The patent segments the data processing workflow into onboard processing (using machine learning models to identify RSOs and extract spectral signatures) and ground-based processing (enriching identification using additional sensors and patterns of life). This segmentation allows critical identification data to be extracted and transmitted onboard, reducing the need to downlink entire hyperspectral data cubes while maintaining high identification accuracy.
Solution Approach 2:
The patent extracts only the essential identification information (RSO classification, spectral signatures, and key features) from the full hyperspectral data cube using onboard machine learning models. This extraction process separates the critical intelligence from the voluminous raw data, allowing transmission of compact identification results rather than complete hyperspectral datasets.
2Quantity of substance
If onboard processing is implemented to reduce data volume, then downlink bandwidth requirements are reduced, but processing capability constraints on spacecraft increase
Solution Approach 1:
The patent implements preliminary action by pre-training machine learning models on ground and downloading them to the spacecraft before missions. These pre-trained models can perform rapid RSO identification and spectral extraction onboard without requiring complex real-time training capabilities, thus reducing onboard processing requirements while still enabling effective data reduction.
Solution Approach 2:
The patent uses simplified copies or surrogate models onboard the spacecraft that replicate the essential identification capabilities of more complex ground-based models. These lightweight model copies enable onboard processing with reduced computational requirements, balancing the need for data reduction with spacecraft processing constraints.
3Measurement precision
If ground-based enrichment processing is used to improve identification accuracy, then RSO characterization is enhanced, but processing time increases
Solution Approach 1:
The patent implements a feedback loop where onboard identification results are transmitted to ground stations for enrichment processing using additional sensors and patterns of life analysis. The ground-based enrichment results are then fed back to update and refine onboard models, creating an iterative improvement process that enhances identification accuracy over time while distributing processing loads between onboard and ground systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for characterizing resident space objects (RSOs) including collecting hyperspectral data of an RSO using a hyperspectral imaging sensor onboard a spacecraft, processing the data using an ML RSO classification and identification model to obtain RSO and component spectra identification data, transmitting the identification data and the hyperspectral data to a ground station, processing the hyperspectral data at the ground station using an image processing algorithm or technique other than the ML-based RSO classification and identification model to obtain enriched RSO and component spectra identification data, optimizing identification of the RSO and spectra in the hyperspectral data using an output of a comparison of identification data and the enriched identification data to obtain optimized identification data; transmitting the optimized RSO and component spectra identification data to a user device; and displaying the optimized RSO and component spectra identification data in a graphical user interface at the user device.