Menu

 

The NASA Task Book
Advanced Search     

Project Title:  Development of a Computer Vision Based Toolbox for Feature Extraction, Analysis, Modeling, and Prediction of Microgravity Data Sets Reduce
Images: icon  Fiscal Year: FY 2025 
Division: Physical Sciences 
Research Discipline/Element:
Physical Sciences: FLUID PHYSICS--Complex fluids
MATERIALS SCIENCE--Materials science
OTHER--Other 
Start Date: 10/01/2022  
End Date: 09/30/2025  
Task Last Updated: 10/05/2024 
Download Task Book report in PDF pdf

Open Science: G-SPACE ASGSR 2023 Conference.pdf 9,575 KB
Open Science: G-SPACE ACCGE 2023 for PSI.pdf 5,771 KB
Principal Investigator/Affiliation:   Cozmuta, Ioana  Ph.D. / G-Space, INC 
Address:  1266 Parkington Ave 
 
Sunnyvale , CA 94087-1559 
Email: ioana@g-space.com 
Phone: 408-391-5912  
Congressional District: 17 
Web:  
Organization Type: INDUSTRY 
Organization Name: G-Space, INC 
Joint Agency:  
Comments:  
Key Personnel Changes / Previous PI: No change in PI. Key Personnel Changes: Dr. Christianna Taylor is responsible for the software development aspects of the platform (replacing Dr. Tibi Stef-Praun). Dr. Brian Motil is a former microgravity PI with over 30 years working at NASA and he will be responsible for the datasets' structuring and processing. Both Dr. Taylor and Dr. Motil joined the team in November 2022.
Project Information: Grant/Contract No. 80NSSC22K1885 
Responsible Center: NASA MSFC 
Grant Monitor: Sansoucie, Michael  
Center Contact: 256.544.5269 
michael.p.sansoucie@nasa.gov 
Unique ID: 15172 
Solicitation / Funding Source: 2021 Physical Sciences NNH21ZDA014N-PSI: Use of the NASA Physical Sciences Informatics System – Appendix G 
Grant/Contract No.: 80NSSC22K1885 
Project Type: Ground,Physical Sciences Informatics (PSI) 
Flight Program:  
No. of Post Docs:
No. of PhD Candidates:
No. of Master's Candidates:
No. of Bachelor's Candidates:
No. of PhD Degrees:
No. of Master's Degrees:
No. of Bachelor's Degrees:
Program--Element: FLUID PHYSICS--Complex fluids
MATERIALS SCIENCE--Materials science
OTHER--Other 
Flight Assignment/Project Notes: NOTE: End date changed to 09/30/2025 per the PI (Ed., 2/13/25).

Task Description: Over two decades of significant effort and resources have been devoted to investigating a broad spectrum of hypotheses in microgravity across the portfolio of physical and life sciences on the International Space Station (ISS). This work has resulted in an impressive amount of data being collected, in particular images and videos. However, much of this image data, to date, remains underutilized because the emphasis continues to be on individual investigations. In this proposal, G-SPACE takes a cross-cutting look at the Physical Science Informatics (PSI) datasets to build a simple computer vision, data analytics, and machine learning tool (ATOM™ toolbox) that would be an enabler to all the PSI users (in particular new PIs) to better interact with the data, standardize data output, and perform insightful analysis on the selected datasets to increase the science readiness of their investigations.

For the past two years, the G-SPACE team has been actively ingesting microgravity data available in the NASA PSI database for the purpose of applying a suite of proprietary algorithms and models from its ATOM™ software platform to extract the delta-to-gravity (™) and utilize it to design and optimize products and manufacturing processes amenable for in-space manufacturing. The platform aims to bridge the gap between microgravity R&D sciences and in-space manufacturing and our team’s hope was that the PSI database would have clean data sets corresponding to ground and flight experiments for ATOM™ to extract the delta-to-gravity™ and focus on microgravity product design and optimization.

Unfortunately the data in the PSI database is simply not ready for this approach. The video data residing in the PSI database has rarely been analyzed to track key features for research, and even if so, it has not been done in an automated manner. The sheer number of images, and the total size of the data set, require an enormous amount of hand-sorting and checking of images and is only available in an unstructured format. This makes it harder for users to find and understand the value that lies in it since to access that information it requires crossing a very high barrier, especially for new Principal Investigators (PIs) who do not usually have previous familiarity.

The current proposal seeks to develop the ATOM™ toolbox, a collection of generic computer vision, data analysis, and machine learning functionalities to help new PIs, as well as existing users, to expand the meaning and interpretation of existing data sets, and to extract heretofore undiscovered knowledge from the PSI database. It will also: (a) enable enhancement of existing data, (b) open up the ability for new researchers to leverage on existing experiments, and (c) help bring the investigations to a faster conclusion.

To develop the ATOM™ toolbox functionalities, the G-SPACE team will look at images and videos only for eight (8) Material Science and two (2) Complex Fluid investigations in the PSI database.

Besides being a powerful tool to extract meaningful information from existing experiments, the ATOM™ toolbox could ultimately provide: (1) the means to guide ISS experiments to make better use of time in microgravity, (2) a mechanism to predict results of future experiments in space for better prioritization and structure in the decision process, and (3) an open door for applications that ultimately create the path towards materials space manufacturing and beyond.

The end products of the two years' effort under this proposal that will be delivered to NASA, to be included in the PSI database, will consist of: (1) a database of structured image/video datasets and a corresponding demo for each of the 10 investigations, (2) the ATOM™ toolbox with a basic application programming interface (API) to allow integration with the PSI database and the G-SPACE ATOM™ platform.

Research Impact/Earth Benefits: The G-SPACE platform:

1. Enables microgravity users to access modern data science and machine learning algorithms specifically tuned to analyze microgravity data, quantify microgravity impact, and establish correlations with environmental variables; 2. Democratize and disseminate microgravity know-how and open up the ability for new researchers to leverage on existing experiments and propose new investigations; 3. Reduce the barrier of entry of new users in the field of microgravity research; 4. Enhance the value of existing data and help increase the science readiness of microgravity investigations; 5. Spark applications towards in-space manufacturing.

Task Progress & Bibliography Information FY2025 
Task Progress: Task Progress Update

Dataset Analysis (10 Experiments):

In several cases, engaging with Principal Investigators (PIs) proved difficult. To compensate, we resorted to published data, making considerable progress. However, validating our results against these publications has been challenging due to the lack of transparency from the authors' side. Despite this, our algorithms successfully processed similar and relevant images, achieving reproducibility rates of 80-90% compared to the published data.

The Physical Sciences Informatics (PSI) data remains unstructured, and piecing together science requirement documentation, hardware details, and flight-related operational records is a cumbersome and time-consuming process. The lack of straightforward documentation presents a significant hurdle.

Many of the provided images are blurry or suffer from poor illumination, which impacts the quality of the analysis and requires further image correction efforts to maintain accuracy.

Platform Development:

The platform is now deployed in beta mode, operational, and has been successfully tested using synthetic data. It has been verified and validated for its intended functionalities. Additional internal testing is ongoing and we will be opening up its testing to a limited set of beta users in Q4 2024.

The only remaining item is the integration with the NASA website, which has been delayed due to the ongoing transfer of NASA Physical Sciences Informatics (PSI) to the GeneLab system.

Bibliography: Description: (Last Updated: 10/23/2024) 

Show Cumulative Bibliography
 
Conference Materials (Downloadable) Cozmuta I, Osan R, Motil B, Bohatel R, Bulugean G. "G-SPACE: an AI-powered microgravity innovation platform." 39th Annual Meeting of the American Society for Gravitational and Space Research, Washington, DC, November 13-18, 2023. , Nov-2023 G-SPACE ASGSR 2023 Conference.pdf (9,575 KB)
Conference Materials (Downloadable) Cozmuta I, Osan R, Motil B, Bulugean G. "Microgravity analytics for material and life science applications." 23rd American Conference on Crystal Growth and Epitaxy (ACCGE-23) and 21st US Workshop on Organometallic Vapor Phase Epitaxy (OMVPE-21), Tucson, Arizona, August 13-18, 2023. , Aug-2023 G-SPACE ACCGE 2023 for PSI.pdf (5,771 KB)
Project Title:  Development of a Computer Vision Based Toolbox for Feature Extraction, Analysis, Modeling, and Prediction of Microgravity Data Sets Reduce
Images: icon  Fiscal Year: FY 2024 
Division: Physical Sciences 
Research Discipline/Element:
Physical Sciences: FLUID PHYSICS--Complex fluids
MATERIALS SCIENCE--Materials science
OTHER--Other 
Start Date: 10/01/2022  
End Date: 09/30/2024  
Task Last Updated: 10/01/2023 
Download Task Book report in PDF pdf
Principal Investigator/Affiliation:   Cozmuta, Ioana  Ph.D. / G-Space, INC 
Address:  1266 Parkington Ave 
 
Sunnyvale , CA 94087-1559 
Email: ioana@g-space.com 
Phone: 408-391-5912  
Congressional District: 17 
Web:  
Organization Type: INDUSTRY 
Organization Name: G-Space, INC 
Joint Agency:  
Comments:  
Key Personnel Changes / Previous PI: No change in PI. Key Personnel Changes: Dr. Christianna Taylor is responsible for the software development aspects of the platform (replacing Dr. Tibi Stef-Praun). Dr. Brian Motil is a former microgravity PI with over 30 years working at NASA and he will be responsible for the datasets' structuring and processing. Both Dr. Taylor and Dr. Motil joined the team in November 2022.
Project Information: Grant/Contract No. 80NSSC22K1885 
Responsible Center: NASA MSFC 
Grant Monitor: Sansoucie, Michael  
Center Contact: 256.544.5269 
michael.p.sansoucie@nasa.gov 
Unique ID: 15172 
Solicitation / Funding Source: 2021 Physical Sciences NNH21ZDA014N-PSI: Use of the NASA Physical Sciences Informatics System – Appendix G 
Grant/Contract No.: 80NSSC22K1885 
Project Type: Ground,Physical Sciences Informatics (PSI) 
Flight Program:  
No. of Post Docs:
No. of PhD Candidates:
No. of Master's Candidates:
No. of Bachelor's Candidates:
No. of PhD Degrees:
No. of Master's Degrees:
No. of Bachelor's Degrees:
Program--Element: FLUID PHYSICS--Complex fluids
MATERIALS SCIENCE--Materials science
OTHER--Other 
Task Description: Over two decades of significant effort and resources have been devoted to investigating a broad spectrum of hypotheses in microgravity across the portfolio of physical and life sciences on the International Space Station (ISS). This work has resulted in an impressive amount of data being collected, in particular images and videos. However, much of this image data, to date, remains underutilized because the emphasis continues to be on individual investigations. In this proposal, G-SPACE takes a cross-cutting look at the Physical Science Informatics (PSI) datasets to build a simple computer vision, data analytics, and machine learning tool (ATOM™ toolbox) that would be an enabler to all the PSI users (in particular new PIs) to better interact with the data, standardize data output, and perform insightful analysis on the selected datasets to increase the science readiness of their investigations.

For the past two years, the G-SPACE team has been actively ingesting microgravity data available in the NASA PSI database for the purpose of applying a suite of proprietary algorithms and models from its ATOM™ software platform to extract the delta-to-gravity (™) and utilize it to design and optimize products and manufacturing processes amenable for in-space manufacturing. The platform aims to bridge the gap between microgravity R&D sciences and in-space manufacturing and our team’s hope was that the PSI database would have clean data sets corresponding to ground and flight experiments for ATOM™ to extract the delta-to-gravity™ and focus on microgravity product design and optimization.

Unfortunately the data in the PSI database is simply not ready for this approach. The video data residing in the PSI database has rarely been analyzed to track key features for research, and even if so, it has not been done in an automated manner. The sheer number of images, and the total size of the data set, require an enormous amount of hand-sorting and checking of images and is only available in an unstructured format. This makes it harder for users to find and understand the value that lies in it since to access that information it requires crossing a very high barrier, especially for new Principal Investigators (PIs) who do not usually have previous familiarity.

The current proposal seeks to develop the ATOM™ toolbox, a collection of generic computer vision, data analysis, and machine learning functionalities to help new PIs, as well as existing users, to expand the meaning and interpretation of existing data sets, and to extract heretofore undiscovered knowledge from the PSI database. It will also: (a) enable enhancement of existing data, (b) open up the ability for new researchers to leverage on existing experiments, and (c) help bring the investigations to a faster conclusion.

To develop the ATOM™ toolbox functionalities, the G-SPACE team will look at images and videos only for eight (8) Material Science and two (2) Complex Fluid investigations in the PSI database.

Besides being a powerful tool to extract meaningful information from existing experiments, the ATOM™ toolbox could ultimately provide: (1) the means to guide ISS experiments to make better use of time in microgravity, (2) a mechanism to predict results of future experiments in space for better prioritization and structure in the decision process, and (3) an open door for applications that ultimately create the path towards materials space manufacturing and beyond.

The end products of the two years' effort under this proposal that will be delivered to NASA, to be included in the PSI database, will consist of: (1) a database of structured image/video datasets and a corresponding demo for each of the 10 investigations, (2) the ATOM™ toolbox with a basic application programming interface (API) to allow integration with the PSI database and the G-SPACE ATOM™ platform.

Research Impact/Earth Benefits: The G-SPACE platform:

1. Enables microgravity users to access modern data science and machine learning algorithms specifically tuned to analyze microgravity data, quantify microgravity impact, and establish correlations with environmental variables; 2. Democratize and disseminate microgravity know-how and open up the ability for new researchers to leverage on existing experiments and propose new investigations; 3. Reduce the barrier of entry of new users in the field of microgravity research; 4. Enhance the value of existing data and help increase the science readiness of microgravity investigations; 5. Spark applications towards in-space manufacturing.

Task Progress & Bibliography Information FY2024 
Task Progress: This document constitutes the Year 1 report of work conducted within the scope of the NASA award “Development of a Computer Vision Based Toolbox for Feature Extraction, Analysis, Modeling, and Prediction of Microgravity Data Sets” (grant 80NSSC22K1885). The work focuses on two key objectives:

1) Create a database of 10 structured image/video datasets from Physical Sciences Informatics (PSI) with a standard file format, clear labeling, and one-to-one traceability to terrestrial data when possible. A brief corresponding ATOM™ demo for each of the 10 investigations will also be developed.

2) Develop the ATOM™ toolbox with a basic application programming interface (API) to allow integration with the PSI database and the G-SPACE ATOM™ platform.

At the end of Year 1, work on six of the structured datasets has been finalized, while work on the remaining four is ongoing. Significant progress has been made in the development of key features for the G-SPACE ATOM™ Toolbox, which includes creating a comprehensive workflow with menus for input, calibration, pre-processing, object detection and extraction, and analytics. An initial executable version has been delivered to the NASA PSI software team. This step is crucial as it establishes a functional framework for G-SPACE, enabling testing of the ATOM™ Toolbox on the NASA PSI website. Additionally, this initiates the integration process within the main NASA PSI website. Specific aspects being addressed include location on the NASA website, menu layout, definition of supporting software architecture, and interfacing with the G-SPACE main platform, as well as outlining maintenance and instruction requirements.

The G-SPACE ATOM™ Toolbox will enable users access to a simple yet universal toolkit via the NASA PSI website. This will empower both existing and new Principal Investigators (PIs) by providing a standardized platform to enhance the initial stages of their analysis. Instead of valuable resources in creating individual analytic tools, researchers could utilize the G-SPACE ATOM™ Toolbox to streamline and accelerate data output, ensuring consistency and insightful analysis. This approach aims to boost the scientific preparedness of their investigations.

Bibliography: Description: (Last Updated: 10/23/2024) 

Show Cumulative Bibliography
 
Abstracts for Journals and Proceedings Cozmuta I, Osan R, Motil B, Taylor C. "An AI predictive platform for microgravity innovation." 23rd American Conference on Crystal Growth and Epitaxy (ACCGE-23) and 21st US Workshop on Organometallic Vapor Phase Epitaxy (OMVPE-21),Tucson, Arizona, August 13-18, 2023.

Abstracts. 23rd American Conference on Crystal Growth and Epitaxy (ACCGE-23) and 21st US Workshop on Organometallic Vapor Phase Epitaxy (OMVPE-21),Tucson, Arizona, August 13-18, 2023. , Aug-2023

Abstracts for Journals and Proceedings Cozmuta I, Osan R, Taylor C, Motil B, Bulugean G. "ATOM™ - an AI predictive platform for microgravity innovation. " 12th Annual International Space Station Research & Development Conference (ISSRDC), Seattle, WA, July 31-August 4, 2023.

Abstracts. 12th Annual International Space Station Research & Development Conference (ISSRDC), Seattle, WA, July 31-August 4, 2023. , Jul-2023

Abstracts for Journals and Proceedings Cozmuta I, Osan R, Taylor C, Motil B, Bulugean G. "Microgravity analytics for material and life science applications." ACS Fall 23 (Meeting of the American Chemical Society), San Francisco, CA, August 14-18, 2023.

Abstracts. ACS Fall 23 (Meeting of the American Chemical Society), San Francisco, CA, August 14-18, 2023. , Aug-2023

Project Title:  Development of a Computer Vision Based Toolbox for Feature Extraction, Analysis, Modeling, and Prediction of Microgravity Data Sets Reduce
Images: icon  Fiscal Year: FY 2023 
Division: Physical Sciences 
Research Discipline/Element:
Physical Sciences: FLUID PHYSICS--Complex fluids
MATERIALS SCIENCE--Materials science
OTHER--Other 
Start Date: 10/01/2022  
End Date: 09/30/2024  
Task Last Updated: 10/04/2022 
Download Task Book report in PDF pdf
Principal Investigator/Affiliation:   Cozmuta, Ioana  Ph.D. / G-Space, INC 
Address:  1266 Parkington Ave 
 
Sunnyvale , CA 94087-1559 
Email: ioana@g-space.com 
Phone: 408-391-5912  
Congressional District: 17 
Web:  
Organization Type: INDUSTRY 
Organization Name: G-Space, INC 
Joint Agency:  
Comments:  
Project Information: Grant/Contract No. 80NSSC22K1885 
Responsible Center: NASA MSFC 
Grant Monitor: Sansoucie, Michael  
Center Contact: 256.544.5269 
michael.p.sansoucie@nasa.gov 
Unique ID: 15172 
Solicitation / Funding Source: 2021 Physical Sciences NNH21ZDA014N-PSI: Use of the NASA Physical Sciences Informatics System – Appendix G 
Grant/Contract No.: 80NSSC22K1885 
Project Type: Ground,Physical Sciences Informatics (PSI) 
Flight Program:  
No. of Post Docs:  
No. of PhD Candidates:  
No. of Master's Candidates:  
No. of Bachelor's Candidates:  
No. of PhD Degrees:  
No. of Master's Degrees:  
No. of Bachelor's Degrees:  
Program--Element: FLUID PHYSICS--Complex fluids
MATERIALS SCIENCE--Materials science
OTHER--Other 
Task Description: Over two decades of significant effort and resources have been devoted to investigating a broad spectrum of hypotheses in microgravity across the portfolio of physical and life sciences on the International Space Station (ISS). This work has resulted in an impressive amount of data being collected, in particular images and videos. However, much of this image data, to date, remains underutilized because the emphasis continues to be on individual investigations. In this proposal, G-SPACE takes a cross-cutting look at the Physical Science Informatics (PSI) datasets to build a simple computer vision, data analytics, and machine learning tool (ATOM™ toolbox) that would be an enabler to all the PSI users (in particular new PIs) to better interact with the data, standardize data output, and perform insightful analysis on the selected datasets to increase the science readiness of their investigations.

For the past two years, the G-SPACE team has been actively ingesting microgravity data available in the NASA PSI database for the purpose of applying a suite of proprietary algorithms and models from its ATOM™ software platform to extract the delta-to-gravity (™) and utilize it to design and optimize products and manufacturing processes amenable for in-space manufacturing. The platform aims to bridge the gap between microgravity R&D sciences and in-space manufacturing and our team’s hope was that the PSI database would have clean data sets corresponding to ground and flight experiments for ATOM™ to extract the delta-to-gravity™ and focus on microgravity product design and optimization.

Unfortunately the data in the PSI database is simply not ready for this approach. The video data residing in the PSI database has rarely been analyzed to track key features for research, and even if so, it has not been done in an automated manner. The sheer number of images, and the total size of the data set, require an enormous amount of hand-sorting and checking of images and is only available in an unstructured format. This makes it harder for users to find and understand the value that lies in it since to access that information it requires crossing a very high barrier, especially for new Principal Investigators (PIs) who do not usually have previous familiarity.

The current proposal seeks to develop the ATOM™ toolbox, a collection of generic computer vision, data analysis, and machine learning functionalities to help new PIs, as well as existing users, to expand the meaning and interpretation of existing data sets, and to extract heretofore undiscovered knowledge from the PSI database. It will also: (a) enable enhancement of existing data, (b) open up the ability for new researchers to leverage on existing experiments, and (c) help bring the investigations to a faster conclusion.

To develop the ATOM™ toolbox functionalities, the G-SPACE team will look at images and videos only for eight (8) Material Science and two (2) Complex Fluid investigations in the PSI database.

Besides being a powerful tool to extract meaningful information from existing experiments, the ATOM™ toolbox could ultimately provide: (1) the means to guide ISS experiments to make better use of time in microgravity, (2) a mechanism to predict results of future experiments in space for better prioritization and structure in the decision process, and (3) an open door for applications that ultimately create the path towards materials space manufacturing and beyond.

The end products of the two years' effort under this proposal that will be delivered to NASA, to be included in the PSI database, will consist of: (1) a database of structured image/video datasets and a corresponding demo for each of the 10 investigations, (2) the ATOM™ toolbox with a basic application programming interface (API) to allow integration with the PSI database and the G-SPACE ATOM™ platform.

Research Impact/Earth Benefits: The G-SPACE platform:

1. Enables microgravity users to access modern data science and machine learning algorithms specifically tuned to analyze microgravity data, quantify microgravity impact, and establish correlations with environmental variables;

2. Democratize and disseminate microgravity know-how and open up the ability for new researchers to leverage on existing experiments and propose new investigations; 3. Reduce the barrier of entry of new users in the field of microgravity research;

4. Enhance the value of existing data and help increase the science readiness of microgravity investigations;

5. Spark applications towards in-space manufacturing.

Task Progress & Bibliography Information FY2023 
Task Progress: New project for FY2023.

Bibliography: Description: (Last Updated: 10/23/2024) 

Show Cumulative Bibliography
 
 None in FY 2023