Space traffic management is complex. Thousands of man-made objects orbit the earth, ranging from space debris to GPS satellites to the International Space Station. Keeping track of all of them is the U.S. Space Force’s space-track.org program. SAIC data scientists asked what insights they could gain from a space-track.org data dump and a few weeks of data analysis, applying Artificial Intelligence and Machine Learning. Learn what they discovered.
SPACE TRAFFIC MANAGEMENT
Home > What We Do > Mission Support > Analytics > Space Traffic Management
The Problem
Advanced Analytics
Data Science
What we could do
These snapshots must be processed to know which object is which and
what it is doing.
Our best observational
data is limited to snapshots over time from our sensor network.
U.S.
SPACE FORCE
Objects are identified and monitored by the
Softball
Softball
Tracked Objects are as small as a
Objects in orbit
23 000
,
THere Are
The Problem
Using these patterns and by comparing historical and trend data, it may be possible to understand uncategorized objects and enhance space traffic tracking fidelity
•
Helps determine when an object is displaying irregular behavior that could mean trouble, such as when the object has struck something and is about to break up
These trends help us generate a pattern of life for space objects
Through advanced analytics of orbit data, we can derive trends for each space object
www.Space-Track.org helps U.S. Space Command facilitate Space Domain Awareness globally
• A safer space environment benefits everyone
Advanced Analytics to the Rescue
The team was given 130 million-plus rows of two-line element (TLE) data files containing orbital elements of space objects (23 .txt files with TLE data spanning 18.6 GB)
Million-plus
TLE Files
130
+
Out of the 45,000 Satellites in the data set we were given five ID numbers as examples of what is considered bad
ID Numbers
5
Data
Solutions
Delivered
The team leveraged open source tool called Cesium that they used to containerize and show the possibility of displaying the results in a 3D viz tool
In the end, the team was able to produce fairly obvious clusters to the human eye in the viz tool, and have some preliminary methods to detect outliers in TLE timelines
The team developed a custom parser to help clean the data. Once cleaned, SAIC data scientists ingested the data to create a dynamic database, all web browser based, from which they sliced and diced off any criteria and queried in near real-time
Data
Solutions
Delivered
•
•
•
•
•
Docker containers enable data science and the visual tool
GitHub Repo delivers tools for ingesting, cleaning, and creating a database that space operators can query dynamically
Visualization illustrates figures, animations, etc.
Process gathers a consensus of clusters through unsupervised clustering
Outlier detection code helps find anomalies
Data
Solutions
Delivered
SAIC’s data science team was given 30 days and 80 funded hours to learn what insights they could derive from a few .txt files of space object data from space-track.org. Little did they know, inputs from satellite tracking experts would be disrupted by the onset of COVID-19. Regardless, the data team was able to clean the data for comparison, identify outliers, create clusters, and visualize the data. If actionable insight equals data plus resources (time, money, expertise, etc.), imagine what SAIC team can do with a data set, a few hours of expert inputs, and a few weeks of funded labor?
Data Science is at the intersection of statistics, computer science, and subject matter experts.
Data Science
•
•
•
•
•
Develop a pipeline to integrate the space_viz tool with the data science process for a more interactive experience
Work with your SME to understand more about what makes up the particular problem they're trying to solve
Build predictive models that answer your particular SME questions
Leverage the additional data in Space-Track about the Sat Cats to gain better clustering or labels to train AI models against
Diagnose malfunctioning satellites - When the algorithm detects a deviation in the pattern of life, we can set up alarms to notify interested parties. This is when using boosters may help
Now imagine the possibilities
What We Could Do
Combining machine intelligence with human intelligence is what leads to gaining new insights. The sum of AI and human expertise is great insight created faster. SAIC is now looking at other areas to explore the art of the possible. Stay tuned to see what we discover next.
Contact us today
Request More Information
or a Demo
TOP
TOP
Space traffic management is complex. Thousands of man-made objects orbit the earth, ranging from space debris to GPS satellites to the International Space Station. Keeping track of all of them is the U.S. Space Force’s space-track.org program. SAIC data scientists asked what insights they could gain from a space-track.org data dump and a few weeks of data analysis, applying Artificial Intelligence and Machine Learning. Learn what they discovered.
SPACE TRAFFIC MANAGEMENT
Home > What We Do > Mission Support > Analytics > Space Traffic Management
These snapshots must be processed to know which object is which and what it is doing.
Our best observational
data is limited to snapshots over time from our sensor network.
U.S.
SPACE FORCE
Objects are identified and monitored by the
Softball
Softball
Tracked Objects are as small as a
Objects in orbit
23 000
,
THere Are
The Problem
Using these patterns and by comparing historical and trend data, it may be possible to understand uncategorized objects and enhance space traffic tracking fidelity
•
Helps determine when an object is displaying irregular behavior that could mean trouble, such as when the object has struck something and is about to break up
These trends help us generate a pattern of life for space objects
Through advanced analytics of orbit data, we can derive trends for each space object
www.Space-Track.org helps U.S. Space Command facilitate Space Domain Awareness globally
• A safer space environment benefits everyone
Advanced Analytics to the Rescue
The team was given 130 million-plus rows of two-line element (TLE) data files containing orbital elements of space objects (23 .txt files with TLE data spanning 18.6 GB)
Million-plus
TLE Files
130
+
Out of the 45,000 Satellites in the data set we were given five ID numbers as examples of what is considered bad
ID Numbers
5
Data
Solutions
Delivered
The team leveraged open source tool called Cesium that they used to containerize and show the possibility of displaying the results in a 3D viz tool
In the end, the team was able to produce fairly obvious clusters to the human eye in the viz tool, and have some preliminary methods to detect outliers in TLE timelines
The team developed a custom parser to help clean the data. Once cleaned, SAIC data scientists ingested the data to create a dynamic database, all web browser based, from which they sliced and diced off any criteria and queried in near real-time
Data
Solutions
Delivered
•
•
•
•
•
Docker containers enable data science and the visual tool
GitHub Repo delivers tools for ingesting, cleaning, and creating a database that space operators can query dynamically
Visualization illustrates figures, animations, etc.
Process gathers a consensus of clusters through unsupervised clustering
Outlier detection code helps find anomalies
Data
Solutions
Delivered
SAIC’s data science team was given 30 days and 80 funded hours to learn what insights they could derive from a few .txt files of space object data from space-track.org. Little did they know, inputs from satellite tracking experts would be disrupted by the onset of COVID-19. Regardless, the data team was able to clean the data for comparison, identify outliers, create clusters, and visualize the data. If actionable insight equals data plus resources (time, money, expertise, etc.), imagine what SAIC team can do with a data set, a few hours of expert inputs, and a few weeks of funded labor?
Data Science is at the intersection of statistics, computer science, and subject matter experts.
Data Science
•
•
•
•
•
Develop a pipeline to integrate the space_viz tool with the data science process for a more interactive experience
Work with your SME to understand more about what makes up the particular problem they're trying to solve
Build predictive models that answer your particular SME questions
Leverage the additional data in Space-Track about the Sat Cats to gain better clustering or labels to train AI models against
Diagnose malfunctioning satellites - When the algorithm detects a deviation in the pattern of life, we can set up alarms to notify interested parties. This is when using boosters may help
Now imagine the possibilities
What We Could Do
Combining machine intelligence with human intelligence is what leads to gaining new insights. The sum of AI and human expertise is great insight created faster. SAIC is now looking at other areas to explore the art of the possible. Stay tuned to see what we discover next.
or a Demo
Request More Information
Contact US to get started