Space: The Final Frontier of Artificial Intelligence
A Conversation with Ibeos-Cubic Aerospace CEO, Abigail Davidson
Some people view the fact that humans have not been to the moon since 1972 as a sign of a lack of advancements in technology relating to outer space. While it’s true that advancement in human space travel has been limited, a different type of intelligence has been making advancements in the cosmos: artificial intelligence (AI). I recently had the opportunity to speak to Abigail Davidson, the CEO at Ibeos-Cubic Aerospace, about some of the advancements in AI technology in space.
Davidson’s background is in electrical engineering, and she used to design satellite electronics for Orbital Sciences Corporation before joining Cubic Aerospace as Director of Programs, then ascended to CEO. Davidson started her career originally at NASA.
Abigail 2nd from right early in her career at NASA
Cubic serves large satellite providers, along the lines of Lockheed Martin and Northrop Grumman, as well as government entities like the Air Force and other smaller clients. Since Cubic is a relatively young and small company, I was curious why clients choose it over a larger corporation. Davidson explained that Cubic is often more agile and less expensive than other options. In addition, it is better at developing certain niche products that aren’t as expensive as traditional space electronics.
Regarding AI and machine learning, one of the most interesting of these niche products is a radiation tolerant payload processing computer that allows Nvidia products that were designed for terrestrial use to be used in space. The space environment is very harsh due to radiation, and it is difficult to use terrestrial graphics processing units (GPUs) and advanced computers in space, but Cubic is trying to change that. However, it will not be easy. According to Davidson, one of the biggest difficulties is cost: it takes millions or sometimes billions of dollars to send satellites into space, so a designer must have high confidence that certain a product will work before sending it up. But if it does, the applications are numerous.
Ibeos-Cubic Aerospace hardware that will be installed on the satellite
Without advanced computing capabilities, data gathered by a satellite must be sent down to earth to be analyzed. This process is lengthy, expensive, and often limited by the transmission capability of a spacecraft’s downlink. However, if the processing could be done in space and only the results were sent down, this process would be a lot faster. This could become a reality if GPUs and advanced computers are put into satellites, as Cubic is trying to do. A satellite could gather image data, analyze the data using AI, generate results, and send those results to earth. With machine learning, these satellites could start discovering trends and patterns that a human can’t. The applications are truly limitless.
Ibeos-Cubic Aerospace's offices in Reston, VA
Cubic Aerospace and companies like it are playing an important role in advancing AI and machine learning applications in space by allowing terrestrial computing products to be used on satellites. I asked Abigail Davidson if she had anything else she wanted to say to our readers, and she responded that the possible AI applications in space go beyond what anyone can currently imagine. Space gives a view of the world few people will ever see, and customers are starting to see the data capabilities this view can provide.
Written By Jack Vasquez, Edited by Rachel Weissman & Alexander Fleiss