Since the beginning, mankind’s exploration of the cosmos has been assisted by the most cutting-edge technology of the time. The needs of pioneering space missions played a major role in the development of computers, cryogenic fuels, high-thrust engines, robotics, radar and sensor technologies and many more. Indeed, not just ‘rocket science’ but also basic and applied sciences, including biosciences, all progressed as a result of humanity’s long-lasting fascination with the Moon and the stars. 

One emerging technological sector which has benefited space missions is Artificial Intelligence (AI). As a quintessential enabling technology, scientists and the industry have proposed widespread uses for AI systems on space missions, some of which have already been implemented. Ranging from AI assistants on board the International Space Station (ISS) to algorithms giving Mars rovers greater autonomy in their activities, various types of AI systems are actively helping humans better understand outer space. 

There are many ways to classify the types of applications AI has in the space arena with the most common methods being to identify them by the area of use (such as in deep space exploration, satellite operation, space systems manufacturing, space science etc) or the type of jobs performed (rovers, image analysis, health monitoring, reducing signal interference etc). However, at its crux, there are two broad spectrum of activities under which the use of AI in space can be classified. 

One is the application of AI to physical robotic systems which are intended to either assist or replace human activity in space and the other is the use of AI for a wide-scope of data-related activities which would then inform human actions. Various types of autonomous activities have been pre-programmed on rovers sent to Moon and Mars or to guide the activity of spacecraft sent to explore the vast reaches of space. However, on more recent missions, such as NASA’s Insight and Mars 2020 rovers, the advances being made in machine learning have been used to try and train these robots to have much greater autonomy in fulfilling their missions. 

The further into space such missions go, the longer it takes for controllers on Earth to be able to communicate with it and assist these spacecraft (whether rovers or orbiters) in making adjustments on the fly. In the case of a Mars mission, such to and fro communication happens at most once or twice every Martian day. Hence, whether it is with respect to their movements, the paths taken to work around obstacles or, as is the case in the Mars 2020 mission, enabling the functioning of the first robotic helicopter on the Red Planet, such autonomous decision-making is central to maximizing the expected output from such missions. 

However, an even more crucial role AI could play on such missions is by enabling the study of that which it was never programmed to. There are inherent limitations to what all a spacecraft is programmed to do before it lifts off from Earth. The greatest of those is that programmers can only program that which they are expecting to encounter and impart the necessary skills to fulfill the mission parameters. Increasingly, though, using machine learning techniques, rovers and orbiters are being trained to recognise and then image objects or events which deviate from the expected picture. The hope is that such systems will be able to detect things which may not have been on the mission mandate at all but could still significantly advance our knowledge of outer space. 

Other cases where AI could get applied to autonomous physical systems includes the examples of humanoids and other assistant robots which could be launched to the ISS or for setting up colonies on the Moon, Mars or elsewhere. Humanoids, such as Russia’s FEDOR have already visited the ISS and AI-powered enhanced versions of similar robots could accomplish various tasks which humans cannot or are too dangerous for humans to do. This could be lifting and transporting heavy cargo or machinery, say on the Moon, or conducting a spacewalk to do repairs under conditions of heavy duress. 

Such technologies have also been trialled for vastly different uses as well such as to monitor the health of astronauts on board the ISS, including their mental faculties, as well as to keep up morale by acting as an interactive companion. AI is also expected to be integrated on key modules of deep space habitats such as the planned lunar orbital platform, the Gateway. One application would be on the Canadarm-3, a robotic system that would both help in auto docking of cargo and crew modules with the Gateway as well as in autonomously conducting repairs and maintenance of the station. In the near future, it is to be expected that AI-augmented nanobots, self driving cars, autonomous resource extraction and storage systems and many more such technologies will get deployed on varying kinds of space missions. 

The second broad area where AI finds application in the space sector relates to its ability to analyze, manipulate and utilize large volumes of data at significantly greater speeds than humans. AI can help in fields like image analysis to identify patterns, narrow down on areas that require greater focus or to quickly cross reference with historic data and point out discrepancies. It can also help satellites perform on-board assimilation of data and thereby direct payloads to zero in on areas that need greater focus without having to wait for ground controllers to communicate that information to it. Such on-board assimilation also opens the door for direct communication of actionable information to the end user instead of waiting for such data to be downloaded, analyzed and then communicated from a central control office. 

AI can also assist in creating predictive models to deal with potential threats such as from space debris, asteroids or meteor showers as well as to track the orbital path of space assets to warn against potential chances of collision. AI-assisted data analysis can also be useful in deciding the path and timing of launches as well as for in-space navigation and maneuvering. With the rise of mega constellations of small satellites resulting in a crowded Low Earth Orbit (LEO), AI tools would be uniquely suited for the requirement of space traffic management and space situational awareness. 

For satellites operators as well, AI can assist in the placement and maneuvering of individual satellites in a constellation to ensure efficient provision of services with the least possible blind spots. Similarly, AI can contribute to the design, placement and positioning of ground stations and antennas to receive the sharpest spectrum of signals from on-orbit systems. At the satellite end of the system, AI can enhance signals being sent to the receivers by reducing the interference coming from other such space assets and choosing the best spectrum through which to send the sharpest signals at different periods of time and at varying distances from the receiving station or user terminal. 

It is evident that AI algorithms whether applied to enhance autonomous capabilities of robots or through machine learning and deep learning techniques to process data accumulated by satellites, have widespread existing and potential uses in the space sector. It can hence be assumed that as the capabilities of AI systems grow, the uses and the efficacy of that use in space missions will also continue to increase. Hence for any nation or private firm looking to develop as a potent player in the outer space arena, investing in having a robust capability in applied AI, along with related fields such as robotics and big data analysis, would certainly be a logical move. 

Indeed, envisioning a long-term future where mankind establishes new colonies out in space would be nearly impossible without artificial intelligence systems. It is perhaps then no wonder that any science fiction story or film which projects large-scale human settlement elsewhere in the universe (or exploration in preparation for colonizing), has also depicted AI systems of varying degrees of intelligence. Whether they are robot assistants like Star Wars’ R2-D2 or Interstellar’s TARS, an interactive artificial consciousness like DC’s Gideon or Marvel’s Jarvis or advanced sentient computers like HAL 9000 (2001: A Space Odyssey) or R.U.D.I (The Jetsons), these stellar characters are certainly likely to provide continued inspiration for the space sector’s use of AI systems into the future.

Disclaimer: The article expresses the author’s views on the matter and do not reflect the opinions and beliefs of any institution they belong to or of Trivium Think Tank and the StraTechos website.