Smarter, farther, more precisely: how does AI change flights into space
We in the Binary District have launched a new course on AI for business . On the course, we talk a lot about the practical application of artificial intelligence in various industries: retail, media, medicine.
But automation helps people not only on Earth. In this article, we have collected examples of how AI can be used (or is already being used) for space exploration.
Under the cut - seven cases: from the prediction of storms on the Sun to the cosmic rob-rescuer.
Scientists from the NASA Frontier Development Lab in collaboration with Microsoft and IBM are working on self-learning systems to predict the strength and scale of solar storms. If successful, they can be used not only to determine the climate of the new planets and their suitability for life, but also to find natural resources.
The laboratory was founded by New Zealand entrepreneur James Parr, inspired by the Obama Asteroid Grand Challenge program in 2013. Parr suggested NASA to work together on a project that introduces advanced AI developments in projects to protect the Earth from asteroids and other hazards. According to Parr, artificial intelligence is the only technology whose benefits astronomers have not yet had time to evaluate.
Artificial intelligence also helps scientists discover new planets. Telescope Kepler, equipped with an algorithm with AI, was launched into space in March 2009 and worked for almost a decade: during this time, astronomers have found more than 2,600 exoplanets using it. For example, at the end of 2017, he helped find the twin of the Solar System by discovering the planets Kepler 80g and Kepler 90i in the Kepler-90 star system in the constellation Dragon.
Render of the Kepler telescope (NASA)
To achieve this result, the researchers trained the algorithm using data from NASA. After studying 15 thousand test signals, the telescope was able to correctly identify the planets in 96% of cases.
If the robot designed Pixar, it would look exactly the same (JAXA / NASA).
The Japanese Aerospace Exploration Agency (JAXA) developed Int-Ball, a remotely controlled drone that captures experiments on board a space station and sends them to Earth
All Int-Ball elements are printed on a 3D printer, it moves with the help of 12 propellers, and it is oriented inside the ISS - along glued pink dots-markers.
A small big-eyed robot (Int-Ball weighs only a kilogram, diameter - 15 cm) facilitates the synchronization of the work of the crew and crew on Earth. Dron assumed several astronaut duties and reduced their workload by 10%. In the future, it is planned that Int-Ball will be able to track product stocks and repair broken ship parts.
The first videos made by Int-Ball
Modern developments with the use of AI are gradually making software and hardware more autonomous, capable of learning. The most anticipated developments in this area are ships capable of independently adjusting the path in relation to the orbit, working on autopilot and landing on the space station.
And AI helps NASA to create concepts of interplanetary landing modules. For example, such
Selecting a site for planting is a complex and multidimensional task. It is necessary that the surface be relatively flat, lit (if the device is solar powered), so that the site is interesting for researchers. In addition, these conditions should be observed on a sufficiently large area - in case the probe does not sit in exactly the intended place, but nearby. In this case, decisions must be made on the basis of incomplete and heterogeneous data on the surface of the planet, collected from several sources.
To solve this problem, scientists have developed a system based on AI, which selects a suitable landing site for the Martian mission. The development is described in detail here.. The technology is based on the theory of fuzzy logic. Unlike ordinary logic, statements can not only be true and false. Fuzzy logic uses terms such as “the statement is true with such a probability” or “the statement is true in such a measure”.
Using data on topography, air, soil composition and other conditions at different points of Mars, the system automatically selects suitable places for rover landing. The program works like this: breaks the surface of Mars into small areas, assigns each a number from 0 to 1 (0 - not suitable for landing, 1 - suitable for landing), groups favorable areas together and offers them to astronomers.
In theory, the program can work in the opposite direction: select the right rover for exploring a particular landscape. The authors of the project hope that soon such algorithms can be used to create autonomous rovers, which are connected to the Earth only in emergency cases. This would make exploring new planets much more efficient and faster.
Search for a suitable location
According to NASA engineer Hiro Ono, autonomous spacecraft are already under development: perhaps Europe, one of Jupiter’s moons, will be the next destination of scientists.
A huge danger to the health of space crew members is radioactive radiation. During the flight, astronauts encounter two types of ionizing radiation at once: solar flares and cosmic rays. Prolonged exposure to such rays destroys DNA chains. The body is able to repair breaks, but during the "repair" errors often occur, leading to mutations.
Scientists from all over the world are conducting joint research in the field of artificial intelligence to continuously monitor the health of astronauts during a flight. The emergence of technology capable of tracking the minimum changes in the condition of crew members, will allow time to take action and avoid serious consequences.
Flying into space is a huge stress for a person, and not only from a physical point of view. Long months spent away from relatives, often without the ability to contact them, is a difficult test for even the most experienced and prepared. Scientists hope that new technologies will help with this. CIMON (Interactive Mobile Satellite Commands) - the first AI-assistant, commissioned by the German Center for Aviation and Cosmonautics. Airbus, together with IBM, developed a virtual assistant to facilitate long flights for crew members.
The robot, similar to a soccer ball, is equipped with several video cameras, microphones, sensors and processors: it communicates with astronauts with their help. Twelve built-in fans allow it to fly in all directions, nod and shake its head.
The main function of CIMON on board a ship or a space station is to give instructions for performing complex tasks or repairing parts of a ship (he can quickly find and organize information). But CIMON is not just an assistant, he also has a social role: to communicate with astronauts during long flights. That is why the developers added the face recognition feature and the “human” element in the form of a wide smile on the screen.
Research in this area is carried out in Russia. The first humanoid rescue robot Fedor (FEDOR - Final Experimental Demonstration Object Research), developed by the Advanced Research Foundation and the Android Technological Scientific and Production Association, may become a member of the crew already in 2021. The robot can drive a car, overcome an obstacle course, use construction tools, navigate the terrain and lift loads weighing up to 20 kg. At the moment it is the only anthropomorphic robot that can crawl on all fours.
Fedor uses two cameras, a thermal imager, a microphone, a GPS and several dozens of lasers to interact with the outside world: this equipment allows him to build a three-dimensional scheme of the environment and more accurately perform tasks. Fedor has four modes of operation: autonomous, supervisory, copying and combined.
Another feature of it is the system of force-moment feedback or sensory communication. The operator uses a special suit to control the robot, and the robot transmits information through the suit back to the operator. Thus, for example, the manager can feel how heavy the load Fyodor raises. In September 2018, Fedor was transferred to Roskosmos , where he would be prepared for flight on the Federation spacecraft.
Over the past few years, space flight has become easier and safer, but in the field of space engineering there remain many unsolved problems. Autopilot ships, social robots and other developments in the field of artificial intelligence can help cope with these problems by making other planets closer and more accessible.
But automation helps people not only on Earth. In this article, we have collected examples of how AI can be used (or is already being used) for space exploration.
Under the cut - seven cases: from the prediction of storms on the Sun to the cosmic rob-rescuer.
1. Predict solar storms and protect from asteroids
Scientists from the NASA Frontier Development Lab in collaboration with Microsoft and IBM are working on self-learning systems to predict the strength and scale of solar storms. If successful, they can be used not only to determine the climate of the new planets and their suitability for life, but also to find natural resources.
The laboratory was founded by New Zealand entrepreneur James Parr, inspired by the Obama Asteroid Grand Challenge program in 2013. Parr suggested NASA to work together on a project that introduces advanced AI developments in projects to protect the Earth from asteroids and other hazards. According to Parr, artificial intelligence is the only technology whose benefits astronomers have not yet had time to evaluate.
2. Discover exoplanets
Artificial intelligence also helps scientists discover new planets. Telescope Kepler, equipped with an algorithm with AI, was launched into space in March 2009 and worked for almost a decade: during this time, astronomers have found more than 2,600 exoplanets using it. For example, at the end of 2017, he helped find the twin of the Solar System by discovering the planets Kepler 80g and Kepler 90i in the Kepler-90 star system in the constellation Dragon.
“It's like looking for a needle in a haystack,” - Chris Schallu, senior engineer at Google AI, one of the project’s researchers, during the opening conference.In October 2018, his fuel tank was empty, and the telescope completed the mission. However, after Kepler had finished hunting for new worlds, astronomers were able to discover another 104 exoplanets , using the data and information he collected from the Gaia telescope.
Render of the Kepler telescope (NASA)
To achieve this result, the researchers trained the algorithm using data from NASA. After studying 15 thousand test signals, the telescope was able to correctly identify the planets in 96% of cases.
3. To make reports from the ISS
If the robot designed Pixar, it would look exactly the same (JAXA / NASA).
The Japanese Aerospace Exploration Agency (JAXA) developed Int-Ball, a remotely controlled drone that captures experiments on board a space station and sends them to Earth
All Int-Ball elements are printed on a 3D printer, it moves with the help of 12 propellers, and it is oriented inside the ISS - along glued pink dots-markers.
A small big-eyed robot (Int-Ball weighs only a kilogram, diameter - 15 cm) facilitates the synchronization of the work of the crew and crew on Earth. Dron assumed several astronaut duties and reduced their workload by 10%. In the future, it is planned that Int-Ball will be able to track product stocks and repair broken ship parts.
The first videos made by Int-Ball
4. Helping vehicles land
Modern developments with the use of AI are gradually making software and hardware more autonomous, capable of learning. The most anticipated developments in this area are ships capable of independently adjusting the path in relation to the orbit, working on autopilot and landing on the space station.
And AI helps NASA to create concepts of interplanetary landing modules. For example, such
Selecting a site for planting is a complex and multidimensional task. It is necessary that the surface be relatively flat, lit (if the device is solar powered), so that the site is interesting for researchers. In addition, these conditions should be observed on a sufficiently large area - in case the probe does not sit in exactly the intended place, but nearby. In this case, decisions must be made on the basis of incomplete and heterogeneous data on the surface of the planet, collected from several sources.
To solve this problem, scientists have developed a system based on AI, which selects a suitable landing site for the Martian mission. The development is described in detail here.. The technology is based on the theory of fuzzy logic. Unlike ordinary logic, statements can not only be true and false. Fuzzy logic uses terms such as “the statement is true with such a probability” or “the statement is true in such a measure”.
Using data on topography, air, soil composition and other conditions at different points of Mars, the system automatically selects suitable places for rover landing. The program works like this: breaks the surface of Mars into small areas, assigns each a number from 0 to 1 (0 - not suitable for landing, 1 - suitable for landing), groups favorable areas together and offers them to astronomers.
In theory, the program can work in the opposite direction: select the right rover for exploring a particular landscape. The authors of the project hope that soon such algorithms can be used to create autonomous rovers, which are connected to the Earth only in emergency cases. This would make exploring new planets much more efficient and faster.
Search for a suitable location
According to NASA engineer Hiro Ono, autonomous spacecraft are already under development: perhaps Europe, one of Jupiter’s moons, will be the next destination of scientists.
5. Track radiation
A huge danger to the health of space crew members is radioactive radiation. During the flight, astronauts encounter two types of ionizing radiation at once: solar flares and cosmic rays. Prolonged exposure to such rays destroys DNA chains. The body is able to repair breaks, but during the "repair" errors often occur, leading to mutations.
Scientists from all over the world are conducting joint research in the field of artificial intelligence to continuously monitor the health of astronauts during a flight. The emergence of technology capable of tracking the minimum changes in the condition of crew members, will allow time to take action and avoid serious consequences.
6. Be a friend
Flying into space is a huge stress for a person, and not only from a physical point of view. Long months spent away from relatives, often without the ability to contact them, is a difficult test for even the most experienced and prepared. Scientists hope that new technologies will help with this. CIMON (Interactive Mobile Satellite Commands) - the first AI-assistant, commissioned by the German Center for Aviation and Cosmonautics. Airbus, together with IBM, developed a virtual assistant to facilitate long flights for crew members.
The robot, similar to a soccer ball, is equipped with several video cameras, microphones, sensors and processors: it communicates with astronauts with their help. Twelve built-in fans allow it to fly in all directions, nod and shake its head.
The main function of CIMON on board a ship or a space station is to give instructions for performing complex tasks or repairing parts of a ship (he can quickly find and organize information). But CIMON is not just an assistant, he also has a social role: to communicate with astronauts during long flights. That is why the developers added the face recognition feature and the “human” element in the form of a wide smile on the screen.
7. Save cosmonauts
Research in this area is carried out in Russia. The first humanoid rescue robot Fedor (FEDOR - Final Experimental Demonstration Object Research), developed by the Advanced Research Foundation and the Android Technological Scientific and Production Association, may become a member of the crew already in 2021. The robot can drive a car, overcome an obstacle course, use construction tools, navigate the terrain and lift loads weighing up to 20 kg. At the moment it is the only anthropomorphic robot that can crawl on all fours.
Fedor uses two cameras, a thermal imager, a microphone, a GPS and several dozens of lasers to interact with the outside world: this equipment allows him to build a three-dimensional scheme of the environment and more accurately perform tasks. Fedor has four modes of operation: autonomous, supervisory, copying and combined.
Another feature of it is the system of force-moment feedback or sensory communication. The operator uses a special suit to control the robot, and the robot transmits information through the suit back to the operator. Thus, for example, the manager can feel how heavy the load Fyodor raises. In September 2018, Fedor was transferred to Roskosmos , where he would be prepared for flight on the Federation spacecraft.
Over the past few years, space flight has become easier and safer, but in the field of space engineering there remain many unsolved problems. Autopilot ships, social robots and other developments in the field of artificial intelligence can help cope with these problems by making other planets closer and more accessible.