NASA Seeks Public's Help to Train AI for Mars Rovers
2021-11-11
LRC
TXT
大字
小字
滚动
全页
1The American space agency NASA is seeking help from the public to identify important images captured by Mars explorers.
2The project is designed to help teach an artificial intelligence, AI, system to recognize different scientific elements seen in the pictures.
3Some of the images were captured by Perseverance, NASA's latest Mars rover, or explorer.
4Others came from the agency's Spirit and Opportunity rovers.
5AI plays an important part in NASA's efforts to process data collected by its spacecraft and exploring vehicles.
6The space agency develops AI algorithms to identify and group elements of scientific interest.
7But the algorithms must be trained by humans in order for them to work.
8That is where private citizens can help.
9NASA says Perseverance can send hundreds of images to Earth on any given day.
10Scientists and engineers examine the images as they search for different geological qualities of the surface of Mars.
11These images can also help guide Perseverance and other rovers operating on Mars.
12But the team on Earth faces a very short timeline.
13NASA said in a recent statement, "After those images travel millions of miles from Mars to Earth, the team members have a matter of hours to develop the next set of instructions."
14Those instructions -- based on what the NASA team sees in the images -- must then be sent back to Perseverance.
15Vivian Sun is a scientist at NASA's Jet Propulsion Laboratory (JPL) in California.
16She helps oversee the daily operations of Perseverance.
17She also advises those who run the public project, called AI4Mars.
18"It's not possible for any one scientist to look at all the downlinked images with scrutiny in such a short amount of time, every single day," Sun said in an online statement.
19"It would save us time if there was an algorithm that could say, 'I think I saw rock veins or nodules over here,' and then the science team can look at those areas with more detail."
20The AI4Mars project aims to build a large, solid set of data created from thousands of images.
21Researchers can use these images to quickly discover interesting geological features.
22It is much easier for other AI researchers to train their algorithms on available pictures of many earthly things.
23But NASA did not have such a massive collection of Mars images to work with.
24Citizen scientists who wish to take part in the AI4Mars project can visit a special website to start.
25There, users will find an online tool that can be used to make lines around features in the images such as sand and different kinds of rock.
26Once these features are marked, a user can choose from descriptions that fit what is in the images.
27NASA says the latest version of the AI4Mars system lets people choose more detailed descriptions.
28Annie Didier is a JPL scientist who works on the Perseverance version of AI4Mars.
29She said creating a strong dataset can serve several important purposes.
30"With this algorithm, the rover could automatically select science targets to drive to," she said.
31It could also permit rovers on Mars to store numerous images and then send back only the ones that scientists are interested in, she added.
32NASA says AI4Mars is the continuation of another project launched last year that used images collected by the Curiosity rover.
33In that project, people were able to outline and mark features like sand and rock from nearly half a million images.
34The result of that effort was an algorithm called Soil Property and Object Classification.
35NASA says that system can correctly identify the features it was trained on about 98 percent of the time.
36I'm Bryan Lynn.
1The American space agency NASA is seeking help from the public to identify important images captured by Mars explorers. 2The project is designed to help teach an artificial intelligence, AI, system to recognize different scientific elements seen in the pictures. 3Some of the images were captured by Perseverance, NASA's latest Mars rover, or explorer. Others came from the agency's Spirit and Opportunity rovers. 4AI plays an important part in NASA's efforts to process data collected by its spacecraft and exploring vehicles. The space agency develops AI algorithms to identify and group elements of scientific interest. But the algorithms must be trained by humans in order for them to work. 5That is where private citizens can help. 6NASA says Perseverance can send hundreds of images to Earth on any given day. Scientists and engineers examine the images as they search for different geological qualities of the surface of Mars. 7These images can also help guide Perseverance and other rovers operating on Mars. But the team on Earth faces a very short timeline. NASA said in a recent statement, "After those images travel millions of miles from Mars to Earth, the team members have a matter of hours to develop the next set of instructions." 8Those instructions -- based on what the NASA team sees in the images -- must then be sent back to Perseverance. 9Vivian Sun is a scientist at NASA's Jet Propulsion Laboratory (JPL) in California. She helps oversee the daily operations of Perseverance. She also advises those who run the public project, called AI4Mars. 10"It's not possible for any one scientist to look at all the downlinked images with scrutiny in such a short amount of time, every single day," Sun said in an online statement. "It would save us time if there was an algorithm that could say, 'I think I saw rock veins or nodules over here,' and then the science team can look at those areas with more detail." 11The AI4Mars project aims to build a large, solid set of data created from thousands of images. Researchers can use these images to quickly discover interesting geological features. 12It is much easier for other AI researchers to train their algorithms on available pictures of many earthly things. But NASA did not have such a massive collection of Mars images to work with. 13Citizen scientists who wish to take part in the AI4Mars project can visit a special website to start. There, users will find an online tool that can be used to make lines around features in the images such as sand and different kinds of rock. 14Once these features are marked, a user can choose from descriptions that fit what is in the images. NASA says the latest version of the AI4Mars system lets people choose more detailed descriptions. 15Annie Didier is a JPL scientist who works on the Perseverance version of AI4Mars. She said creating a strong dataset can serve several important purposes. "With this algorithm, the rover could automatically select science targets to drive to," she said. It could also permit rovers on Mars to store numerous images and then send back only the ones that scientists are interested in, she added. 16NASA says AI4Mars is the continuation of another project launched last year that used images collected by the Curiosity rover. In that project, people were able to outline and mark features like sand and rock from nearly half a million images. 17The result of that effort was an algorithm called Soil Property and Object Classification. NASA says that system can correctly identify the features it was trained on about 98 percent of the time. 18I'm Bryan Lynn. 19Bryan Lynn wrote this story for VOA Learning English, based on reports from NASA. Ashley Thompson was the editor. 20We want to hear from you. Write to us in the Comments section, and visit our Facebook page. 21__________________________________________ 22Words in This Story 23artificial intelligence (AI) - n. an area of computer science that deals with giving machines the ability to seem like they have human intelligence 24algorithm - n. a set of steps that are followed in order to solve a mathematical problem or to complete a computer process 25scrutiny - n. careful examination in a critical way 26vein - n. a narrow water channel in rock or earth 27nodule - n. a small, rounded piece of matter that is different from its surroundings 28feature - n. a typical quality or important part of something 29automatic - adj. something controlled using machines and not people 30select - v. to choose 31timeline - n. a line that shows the time and order of the way things happened 32outline - v. to draw, trace, or define the outer edge or shape of something