The Startup
Published in

The Startup

How to Conquer Space Using AI

A guide on how AI is and will continue to revolutionize the space industry


Current Uses


Space Exploration

Asteroid Detection

Satellites-The Data Crunch


Predicting Hurricanes

Debris Problem

Habitat Building

Potential uses


Reducing Battery Usage

More Potential Uses For AI In Space

Key Takeaways

When you think of AI in space, you probably think of one of these characters.

C-3PO and R2-D2 from Star Wars Credit:

Before modern computers were even made, science fiction was planting ideas in our brains of AI in space. Unfortunately, the AI we have in space isn’t that cool. Although we do need AI in space because space is a challenging place. Some tasks require extreme precision that humans aren’t capable of achieving.

Satellites produce about 16 images every second that are low quality and can’t be processed as well with regular code. The satellites have poor communication. Satellites make and need to process 150 terabytes of data every day! That data won’t be useful tomorrow. Who would want the weather from yesterday? Also more importantly, if you want better photos or data on the object the satellite is looking at. If you don't process it quickly enough; you will have to wait until the satellite orbits all the way around to get a chance to get the data or photo again. Even worse, you could have missed it because the thing you want has moved or has changed. Even more important - there is an asteroid coming towards the Earth! Unfortunately, the alert was too late and only the billionaires survived and have gone to Mars. That is more important than the weather people!

Current Uses


Cobots or collaborative robots are robots that are built for direct contact and interaction with humans like a robot dog or a robotic vacuum. There have been a surprising amount of cobots in space.

CIMON 1&2(Crew Interactive MObile companioN)

DLR project manager Christian Karrasch and Airbus project manager Philipp Schulien posing with CIMON in front of the Vehicle Assembly Building of NASA’s Kennedy Space Center. Image Credit: Wikipedia

CIMON was made by IBM, AIRBUS and the DLR (German Aerospace Center). The original CIMON was first proposed in 2016 and went to the ISS in 2018 for 14 months. CIMON 2 went up to the ISS on December 5th, 2019 and it is scheduled to stay there for 3 years. At its core, it’s a science experiment to demonstrate how cobots like CIMON could benefit the ISS crew and other people in space. The benefits of CIMON are reducing crew stress, improving ISS security and helping the crew perform tasks more efficiently. CIMON is a spherical AI assistant that has a diameter of 32cm, weighs 5kg and has 3 hours of battery life. CIMON can speak, fly, help with some tasks, play music, see, listen and give alerts to the crew. CIMON also uses the IBM cloud for data security. CIMON is compatible with AR, is fully built using 3D printing, has facial recognition to recognize people and facial expressions.

Int-Ball (Jem Internal Ball Camera)

JEM Internal Ball Camera, Int-Ball, image credit: JAXA/NASA
JEM Internal Ball Camera, Int-Ball Image Credit: JAXA/NASA

The Int-Ball, also know as the JEM Internal ball camera is an experiment and has been on the ISS since he flew up on June 4th, 2017. Int-Ball was made by JAXA (Japan Aerospace Exploration Agency). The Int-ball is 15cm in diameter, weighs 1kg, has a battery that lasts about 2 hours and the outer body is made of 3D printed material. The Int-Ball can fly autonomously using 12 small fans and 3D markers. It can also be flown and monitored by the JAXA ground crew. It can take photos autonomously in HD 1280 x 720 pixels and FHD 1920 x 1080 pixels. The Int-Ball helps the ISS crew by taking photos and recording videos for scientific or public purposes, which usually takes up 10% of the ISS crew's working time.


Kirobo is Japanese for hope. Image Credit: ABC News

The purpose of Kirobo was to see how humans and cobots could interact in space. Kirobo was the first companion robot with AI to go to space. Kirobo arrived on the ISS on August 10th, 2013 and stayed for 18 months. He is about 34 cm tall, 18 cm wide and 15 cm deep. It weighs about 1 kg and speaks Japanese. Kirobo was made by Dentsu, the University of Tokyo’s Research Center for Advanced Science and Technologies, Robo Garage, Toyota and JAXA(Japan Aerospace Exploration Agency). Kirobo’s abilities included voice and speech recognition, natural language processing, speech synthesis, telecommunications, facial recognition and video recording.

Robonaut (R1 & R2)

Two Robonaut 2s in july 2009
R2s in July 2009 Image Credit: Wikipedia

Robonaut is a humanoid robot -which means he is designed to act and look like a human. Robonaut was made by GM and NASA and can be controlled remotely or “think” on its own. Robonaut was made to work along with other astronauts and his purpose was to see how Robonaut or cobots like Robonaut could help other astronauts. Robonaut’s designers have ideas for sending cobots like Robonaut to Mars or the Moon. Robonaut 1 never made it to space. Robonaut 2 was faster, smaller, more dexterous, and had a deeper and wider range of sensing. Both Robonaut and Robonaut 2 had multiple lower bodies; but Robonaut 2 had more. Robonaut 2 had a zero-g leg, could use Centaur 2 as a lower body, climbing legs, and more. Robonaut 2 went to the ISS on the STS-133 on February 24, 2011, and in May 2018 Robonaut returned to earth.


Astrobee is three different cobots: Honey, Queen Bee and Bumble. The Astrobee cobots are cubes shaped cobots and 31.75cm wide. The Astrobee robots can fly autonomously using small fans. Their purpose is to help the astronaut perform maintenance tasks on the ISS.

This is just the start of some of the space cobots powered by AI, assisting astronauts in space in many different ways.

Space Exploration

The curiosity mars rover’s selfie
The curiosity mars rover’s selfie

Right now, there is almost none space exploration using AI. Our favourite Mars rover, Curiosity, uses AI to gather data efficiently and will use AI to avoid the Spirit Mars rover…BIG mistake. First, a little backstory on the Curiosity rover. The Curiosity rover has been on Mars since 2012, or for 3005 days. His mission is to explore Mars. OK. Now that’s done. The Curiosity Mars rover uses the AEGIS software, which allows the Curiosity rover to autonomously select the proper rock and soil targets for analysis. After implementing the AEGIS software into the rover’s laser detection system, the performance in selecting the appropriate target was higher than 93% as opposed to the 24% before using a random selecting approach; but, I imagine any sophisticated system would be better than that. The Curiosity Mars rover also is going to use AI for navigating around sand traps, sharp rocks and other obstacles on the red planet, and avoid the Mars Spirit rover mistake of hitting a sharp rock and dying! Right now there is an online tool called AI4Mars which allows you, YES, YOU!, to help label terrain surrounding the Curiosity Mars rover; so, it can avoid the mistake of its fellow Mars rover. Another use of AI in space exploration is in NASA's collaboration with Google for the Kepler mission to find exoplanets using data on the brightness of the star of exoplanets’ solar system.

Asteroid Detection

A 3D model of asteroid Eros. Credits: NASA’s Scientific Visualization Studio
A 3D model of asteroid Eros.
Credits: NASA’s Scientific Visualization Studio

Usually, astronomers have to catalogue asteroids manually using math and telescopes, to find out the shape, size, spin rate and trajectory. This process can take months to complete manually. With a new neural network algorithm, they can render an asteroid shape, size, spin rate and general trajectory in 4 days. A team in the Netherlands has developed a neural network algorithm called “Hazardous Object Identifier” that can identify asteroids that are on a collision course with earth. This algorithm has found 11 new asteroids that are on a collision course with the earth that was previously unknown and are larger than 100 metres in diameter. The algorithm also looked at asteroids that will come within 4.7 million miles, which have a pretty slim chance of hitting the earth. The algorithm has a 90.99% accuracy, which is ok, but means it misses about 1/10 asteroids. Another team from the Aerospace Corporation’s Artificial Intelligence Analytic and Innovation Department developed an AI algorithm using 100 terabytes of data provided by the Catalina Sky Survey team. The algorithm called NEO AID (Near-Earth Object Artificial Intelligence Detection) is being tested at the Catlina Sky Survey to find asteroids and prioritize them. It has increased performance by 10% and the aerospace engineers there predict the results will be even higher. The boost comes from prioritization, the algorithm puts important asteroids like asteroids coming towards earth on the top of the stack for analysis and unimportant asteroids like asteroids that are going to miss earth at the bottom of the stack to analyze. They hope to use it to find ways to redirect asteroids and they predict soon it will be,

AI getting the job done and humans are helping.

Satellites-The Data Crunch

Satellites provide us with valuable information but one thing satellites do during this process is created lots and lots of data and images, about 150 terabytes of data a day is produced by satellites and one thing AI is really good at is processing lots of data very quickly. So it’s no surprise that AI is being used to analyze data in satellites on:

  • Crops’ biochemical and biophysical state
  • Autonomous Cars
  • Soil moisture
  • Satellite images
  • Mobile locations
  • A lot of the IoT (Internet of Things)
  • And many, many, more

Here are some more things AI has does with satellites in more depth:

Managing Satellites

NASA has lots of satellites in orbit. They help NASA study the oceans, land and atmosphere. Image Credit: NASA

Satellites are easy to use, just shoot them into orbit and use them. WRONG! One of the major uses of AI and ML for satellites is making satellites automated because there is a lot of human work that used to go into satellites. With new AI and ML, algorithms satellites can make a course correction automatically to avoid debris and other satellites, be automatically monitored and maintained, automatically communicated with ground crew and automatically respond to stimuli and sensors. Airbus has also been using Google’s open-source AI language, Tensor Flow to monitor the health of Airbus satellites in orbit to make their satellites even better. IBM even has a project called KubeSat which allows satellites to communicate with each other creating satellites swarm intelligence in space.

Mobile Networks

Using AI with satellites can easily give people the internet in underserved areas and places where cables, fibres or mobile networks will not be available without having to launch basically a ring of satellites surrounding the earth. Also at Hughes, they use AI to prevent undesired network behaviours like no phone service or slow phone service. The algorithm prevents about 70% of undesired network behaviours.

Satellites Image Enhancement

Satellite images are extremely high resolution but what if you wanting to see something even more clear. SuperRes is an AI algorithm that automatically enhances the quality of satellite images. This is very useful because it allows lower cost, lower quality and lighter image sensors for satellites. Airbus has also developed an AI algorithm using Tensor Flow and billions of square kilometres of imagery dating back to 1986 for automatic cloud detection (like literal clouds), removing the manual check that was previously needed before image delivery. This algorithm can now detect objects like cars, boats or planes.

Monitoring Forest Health

Our trees are not doing too great and they provide us with vital oxygen. So, using satellite imagery and AI, monitors forests to find forest disturbance, insect plaque, drought, etc. This is useful because we will be able to find and treat problems in our trees all over the world, very easily, cost-efficiently and accurately.

Image Credit:

Detecting Anomalies

LatConnect 60 has made an AI algorithm that can detect real-time anomalies such as a major temperature change in the ocean and takes a picture with a timestamp, coordinates or other types of data.


Relativity Space extremely large 3D printer that uses AI to print Rocket and Satellite parts Image Credit: Relativity Space

AI doesn’t just help space tech in space but also during its creation as well. AI is being used by many people like Lockheed Martin to analyze the manufacturing process to improve it, analyze the satellite, rocket, etc in space to improve future models, analyze the work performed to properly ensure the process has been done correctly, perform tests and speed up the manufacturing process. Relativity Space is an example of a rocket manufacturing company that uses AI in these ways to optimize their manufacturing in every way possible and to let themselves create customized solutions into reliable flight parts and to reduce the number of parts.

Predicting Hurricanes

Using satellite imagery, AI and info found by Scientists at NASA’s JPL (Jet Propulsion Laboratory), IBM Watson studio made an algorithm that can predict a hurricane’s path and the chance of it become more severe. It does this by looking at the amount of rainfall in the core of the hurricane, the amount of ice in the clouds and the temperature of the air flowing away from the eye of the hurricane. The algorithm can predict about 60% better than humans that a hurricane’s winds would increase to a minimum of 56km per hour within 24 hours and 200% better than humans that the winds speeds would go up to 64km per hour within 24 hours.

Debris Problem

Every time we lose something in space or something blows up to space; we create debris which moves at thousand of metres a second making it harder for the next rockets, satellites, etc to go and stay in orbit. IBM has developed a system called SSA(Space Situational Awareness) which is an AI system that can predict and track debris to make sure rockets, satellites and anything else doesn’t get hit and destroyed by some debris that is even as small as a golf ball.

Habitat Building

The AI Space Factory team posing in front of their mars habitat at the NASA 3D print habitat challenge

In 2033, NASA plans to send people to Mars, and in 2022 Elon Musk wants to send people to Mars but the astronaut needs a place to live once they get there. You can’t fly a house to Mars nor the heavy building materials. Also, the habitat on Mars must be airtight, be able to survive dust storms that can last for months, survive temperatures ranging from -129C (200F) to -46C(-50F) and any cosmic debris. Instead of building it when we get there, we can 3D print a habitat for the astronaut using AI before they reach Mars. AI Space Factory is a company at the forefront of habitat building for Mars. Their habitat is 3D printed and made from basalt which can be found on Mars and a plastic polymer made from corn which NASA wants to grow on Mars to reduce the number of materials that need to be shipped. Also, it’s an egg-shaped structure designed by a generative design AI which allows them to create the optimal structure that is extremely strong, uses minimal material, has optimal thermal insulation, etc.

Potential Uses

There are so many potential uses for AI in space so I'm not going to talk about all of them. I know, I know it would be so much fun to make and read but I don't have that kind of attention span but apparently you do because you are still here. Anyways.


Space-Based Battle Intelligence
Satellites can even be used to deliver intelligence to people on the front lines of war or maybe, behind a computer being used for warfare. Usually, data processing and exploitation happens on the ground and satellites download the data but Raytheon Intelligence & Space, are working on systems where the satellites become the data collected, data exploiter and the data sender.

SmartSat is using AI on satellites to improve cybersecurity using heuristic patterns to detect anomalies. which improves cybersecurity on-board with automatic updates as new threats emerge.

Reducing Battery Usage

AI can even save power for the ISS or satellites. Of course, this doesn’t just apply to space technologies, but it’s extremely useful for space technologies because any weight that can reduce can make it way easier for the initial launch. Also, more efficient energy usage will allow people access to more energy in space which will open previously closed doors for space technologies.

More Potential Uses For AI In Space

There are so many ways AI could transform space but we have to be careful because our AI model is only 99% effective in space that 1% means people could die.

Key Takeaways

  • Space is an extremely harsh environment.
  • AI can revolutionize the field of space like crazy and it already is!
  • Space exploration means lots of data and AI is good with lots of data.
  • Soon we could see AI replacing human astronauts altogether because it minimizes the risk of life loss, is cheaper, more efficient, lighter and makes fewer mistakes.

If you like this article you will probably like my other ones so consider following me on medium and well you’re doing that follow me on Twitter, Linked-in and sign up for my newsletter




Get smarter at building your thing. Follow to join The Startup’s +8 million monthly readers & +756K followers.

Recommended from Medium

Down the Rabbit Hole: June 14th, 2021

A First Glimpse into the Bot Builder Landscape in 2021

Achieving Trustworthy AI Depends on the Consolidation of These 3 Pillars

Chatbots Are Not Dead (Here’s Why)

AI Isn’t Stealing Jobs, It’s Making Them More Human

“More than 1 in every 3 organizations have suffered from AI bias”

Beetroot and Parashift Jointly Offer AI-Based Solutions for Digital Document Processes out of the…

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Thomas Lawrence

Thomas Lawrence

I’m a curious 17-year-old. I’m interested in QC, AI and many other things.

More from Medium

What is Web3 Liquidity Aggregation?

The Artemis Plan: Humankind’s Return to the Moon

Just Do It: Mandate Ransomware Reporting

Quantum Computing’s potential for Automotive