With the help of artificial intelligence, can exploring space be 10 times more efficient?

Artificial intelligence (AI) in space exploration is gaining momentum, reports SingularityHub. In the coming years, new missions look like they could be greatly aided by AI as we travel to comets, moons, and planets, and explore the possibility of mining asteroids.

Artificial intelligence (AI) in space exploration is gaining momentum, reports SingularityHub. In the coming years, new missions look like they could be greatly aided by AI as we travel to comets, moons, and planets, and explore the possibility of mining asteroids.

Leopold Summers, director of the Advanced Concepts and Research Office at the European Space Agency (ESA), said in an interview: “AI has changed the game, making scientific research and exploration more efficient. AI not only makes this It doubled the efficiency, but increased it by a factor of 10.”

Examples abound

The application of AI in space exploration is much older than many people think. AI is already playing an important role in studying our planet, solar system and universe. As computer systems and software evolve, so do the potential use cases for AI.

The Earth Observer 1 (EO-1) satellite is a good example. Since its launch in the early 2000s, its onboard AI system has helped optimize analysis and response to natural disasters such as floods and volcanic eruptions. In some cases, the AI ​​has even been able to let the Earth Observer 1 satellite start taking images before ground crews are aware of the accident.

Other satellite and astronomical examples abound. In the second Palomar Sky Survey, the Sky Image Cataloging and Analysis Tool (SKICAT) has assisted researchers in classifying celestial objects found. The Palomar Sky Survey was designed to classify tens of thousands of images of objects taken at low resolutions, far beyond the capabilities of humans. Similar AI systems have helped astronomers identify 56 new, possible “gravitational lenses” that play a key role in dark matter research.

The ability of AI to scour large volumes of data and find correlations will become increasingly important as it makes the most of existing data. The European Space Agency’s ENVISAT generates around 400 terabytes of new data each year, but that pales in comparison to the Square Kilometre Array, which produces about the same amount of data per day that is currently available on the Internet.

AI helps land on Mars

AI is also used for orbit and payload optimization. Both missions are important steps for NASA’s next Mars 2020 Rover mission, which will land on Mars in early 2021. AI, known as AEGIS, is already present on NASA’s current Mars rover, a system that helps cameras automatically aim and select subjects to investigate. However, next-generation AI systems will be able to control vehicles, autonomously assist with research selection, and dynamically schedule and execute scientific missions.

Throughout his career, John Leif Jorgensen from DTU Space, Denmark, has designed numerous devices and systems that are used on more than 100 satellites. Jorgensen is a team member of the autonomous science instrument PIXL on the Mars 2020 Rover, which makes extensive use of AI. The aim is to investigate whether stromatolite-like life forms exist on Mars.

“PIXL’s microscope is mounted on the detector’s arm and needs to be placed 14mm from what we want to study,” Jorgensen said in an interview. “This is thanks to several cameras mounted on the detector. This listens to It might be simple, but the handover process and pinpointing the exact position of the arm is very difficult, like identifying a building from a street photo taken on a roof. However, this is something that AI is particularly well suited to do.”

AI also helps PIXL operate automatically at night, constantly adjusting as the environment changes. On Mars, day and night temperatures can vary by more than 100 degrees Celsius, which means that the distance to the ground beneath the rover, camera, robotic arm and the rock being studied is constantly changing. “AI is at the heart of all of this work and helps to more than double the efficiency,” Jorgensen said.

Mars first, then satellites

Mars is likely to be far from the ultimate destination for AI in space exploration. Jupiter’s moons have long fascinated scientists. In particular, Europa, which may have a subsurface ocean buried under about 10 kilometers of ice. It is one of the most likely places in the solar system to find life, other than Earth.

While the mission may be completed sometime in the future, NASA’s current plan is to launch the James Webb Space Telescope into orbit about 1.5 million kilometers above Earth in 2020. Part of the mission will involve an AI-powered autonomous system that oversees the full deployment of the telescope’s 705kg mirror.

The distance between Earth and Europa, or the distance between Earth and the James Webb telescope, means that communications will be delayed. This, in turn, makes it imperative for astronauts on space missions to be able to make their own decisions. The example from the Mars rover project shows that communication between the Mars rover and Earth requires a 20-minute delay due to the distance. The Europa mission could have even longer communication delays.

These two tasks illustrate, to varying degrees, one of the most significant challenges currently facing the use of AI in space exploration. There is often a direct link between the performance of AI systems and the amount of data they receive. The more data there is, the better the AI ​​system will perform. But we don’t have a lot of data to train such a system to anticipate what challenges might be encountered in a mission to a place like Europa.

Computing Power is the second challenge. The painstaking and time-consuming approval process and radiation risks mean that in the near future, the computer in your home may be more powerful than anything that goes into space. The 200MHz processor, 256MB of RAM, and 2GB of memory sounds more like the Nokia 3210 than the iPhone X, but it’s actually the “brain” of the next-generation space probe.

Private enterprise takes off

Private companies are helping to break through these limits. CB Insights counted 57 startups in the space sector, spanning various fields including natural resources, consumer tourism, research and development, satellites, spacecraft design and launches, and data analytics. David Chew, an engineer at Japanese satellite company Axelspace, explains how private companies can improve the efficiency and reduce costs of space exploration.

David Chew said in an interview: “Many private space companies are using fall-back systems, looking for ways to use what traditional companies think of as non-space-grade components and systems. By implementing fall-back operations, and using AI, it is possible to Integrate and use lower cost components without increasing the risk of failure.”

Transforming our future home

In the more distant future, such feats as modifying the Martian environment are waiting to be realized. These projects to turn other planets into Earth-like environments would not have been possible without the help of AI. Autonomous vehicles have begun to “transform” on Earth, with BioCarbon Engineering using drones to plant 100,000 trees in one day. Drones first survey and map an area, then algorithms decide the best location for trees before a second wave of drones do the actual planting.

As is the case with exponential technologies, the potential for synergies and convergence is enormous. Like AI and robotics, or quantum computing and machine learning. Why not put an AI-powered robot on Mars and make it a teleoperation target for geoscientists? Suffice it to say that we are already in the early stages of using virtual reality (VR) and augmented reality (AR) systems that take data from Mars rovers to create a virtual landscape where scientists can walk around and make decisions about the rover The next exploration goal.

One of the biggest benefits of AI in space exploration may not have much to do with its actual function. David Chew believes that in just 10 years, we could discover the first mining-capable asteroids in the Kuiper Belt with the help of AI.

He said: “What I think AI has contributed to space exploration is that it has opened up a range of new possible industries and services that will have a more immediate impact on human life on Earth. It has become a resonant Industry, has a real impact on people’s daily lives. To the extent that space exploration has become part of the way people think, the border between Earth and the solar system has become less important.”

The Links:   LM64P807 MCC200-16IO1