Categories
Cosmic Space

Intel office enlists AI to analyze satellite images

U.S. intelligence officials today launched a program to develop new satellite image analysis tools that use machine learning and other tricks of the artificial intelligence trade.

In a news release, the Intelligence Advanced Research Projects Activity announced that contracts for the Space-based Machine Automated Recognition Technique program, or SMART, have been awarded to teams led by BlackSky Geospatial Solutions, which has offices in Seattle as well as Herndon, Va.; Accenture Federal Services, with offices in Arlington, Va.; Kitware, headquartered in New York; and Systems & Technology Research in Woburn, Mass.

Today’s announcement didn’t disclose the amounts of the awards. The work is to be done in three phases over the course of four years.

SMART aims to automate the quantitative analysis of space-based imagery to search for natural or human-caused events of interest. Such events could range from construction activity and battle damage to wildfires, flooding and earthquakes.

Automated analysis is becoming more crucial as the flow of Earth imagery from satellites builds to a flood. SMART systems will be designed to look at thousands of images from multiple sources, search for specific activities of interest, and track those activities over time.

“Current manual exploitation methods do not scale well with the data volumes we’re receiving, and there’s the problem of simultaneously analyzing data from past, current and future space-based systems,”  IARPA program manager Jack Cooper said in today’s news release. “SMART innovations in data fusion and machine learning techniques will enable automated broad area search at unprecedented temporal resolution and area coverage.”

BlackSky plans to expand upon its Spectra AI geospatial data analysis platform to develop an automated system for monitoring large-scale construction of critical infrastructure such as military bases, stadiums, campuses, dams and airports.

“This is a tremendous breakthrough in unsupervised learning for our Spectra AI platform and an unprecedented step toward the future of global monitoring,” BlackSky CEO Brian O’Toole said in a news release. “The IARPA SMART program is a natural fit for BlackSky, given our deep expertise in geospatial analytics and our proven ability to deliver first-to-know insights.”

Under its contract, BlackSky will create open-source, supervised and semi-supervised learning algorithms to recognize data patterns specific to large-scale construction projects. The effort will leverage the resources of the NASA/USGS Landsat constellation and the European Space Agency’s Sentinel constellation.

BlackSky’s Spectra platform has been supported by several Pentagon awards in the past, including an Air Force contract issued last July to track the effects of the coronavirus pandemic on military interests worldwide. BlackSky is also one of the satellite ventures under consideration for providing imagery to the U.S. government through a long-term arrangement with the National Reconnaissance Office.

BlackSky currently has six of its Global satellites in orbit, and aims to boost that Earth-observing constellation to 16 operational satellites by the end of this year. The company, which could be considered the corporate offspring of Seattle-based Spaceflight Industries, is already working on the next generation of Global satellites.

By Alan Boyle

Mastermind of Cosmic Log, contributing editor at GeekWire, author of "The Case for Pluto: How a Little Planet Made a Big Difference," president of the Council for the Advancement of Science Writing. Check out "About Alan Boyle" for more fun facts.

Leave a Reply