How neural, or learning network makes astronomers work easier?
Neural networks are powerful tools in many fields. The term "neural networks" means learning networks, and when astronomers are operating with this kind of system they are using databases, where is stored data from the universe. One type of data is a simple image of the sky, which are showing things like the position of asteroids at a certain time. And when the system detects some new white spots in those images, that thing might mean new objects in our solar system.
But the thing where the neural network is the best is that the system can follow large entirety. So, in this case, artificial intelligence follows movements of the thousands or even billions of particles in the asteroid belt. The problem in that area of our solar system is that there are so many particles, that following each of them is difficult, and the human observer cannot even make that thing.
One of the biggest problems in the cases, where the movements of individual targets are observing in the asteroid belt is that the images of a certain area must take in a certain time. In this version of the research, the system takes pictures simultaneously after a certain period. And the purpose of this process is trying to find new spots from those images.
The problem with large telescopes is that they are observing only the small part of the sky. The second problem is that those giant telescopes have only limited observation time. When the smaller telescopes are used to find out interesting targets, the system can call larger telescopes to confirm that thing.
The smaller telescopes can observe larger areas, and they are seeing multiple targets. So if the new spot is noticed, the bigger telescope would call to take a sharper image of that spot. That would share the stress of those telescopes more smoothly.
So in this case, if the new spots in some areas are found, the system can ask bigger telescopes to take an image of that object. Then artificial intelligence can compare the image of the object with other images, what are portraying asteroids. The system is similar, which is used in face recognition in the mobile telephones, and it compares the shape and color of the object and the form of impact craters with the known asteroids, and that will tell is the asteroid already known or the new acquaintance.
The problem with the calculating trajectories of the asteroids is that the Earth is a rotating planet. So the observatories must cooperate in this kind of mission. That means that the observatories are connected by using the internet, and when the object lays in the horizon in the position of the first observatory, the second one continues its mission. The images are stored in databases like film, and artificial intelligence can follow the movements of those asteroids.
But when artificial intelligence is calculating trajectories of those objects it must define the beginning point for the series of observations. Then the system must start to take photographs of those asteroids simultaneously, and in that process, the observatories can connect worldwide by using the internet. That means the observatories around the world can take images from the same area and in this process, the system must make sure that the asteroids that are under surveillance are the same all the time.
Sources:
Comments
Post a Comment