Space probes must gain artificial intelligence and autonomy while conducting an alien search … [+]
Space probes send too much data. In fact, scientific instruments that study the geochemistry of planetary bodies, and search for extraterrestrial life, generate increasing amounts of data.
It is slow, expensive to send into space, and the volume overwhelms astronomers. So why not incorporate artificial intelligence into space probes and allow them to decide what data to disturb humans with?
In short, have them call home only when they have found something that they consider really important.
That’s the focus of a presentation made this week by two scientists from NASA’s Goddard Space Flight Center at the Goldschmidt Geochemistry conference, which will take place virtually June 21-26, 2020.
It could be a “visionary step in space exploration” according to one of the researchers.
The intention is to install these new intelligent systems on space probes, beginning with the ExoMars mission in 2022, which can identify the geochemical signatures of life from rock samples before sending back only relevant data back to Earth.
A functional prototype of the ExoMars rover at the Airbus Defense Space facilities on February 7, 2019 in … [+]
Allowing these intelligent systems to choose both what to analyze and what to tell us on Earth will, scientists say, overcome the severe limits on how information is transmitted over great distances in space.
“Smart space probes” could be of fundamental importance as the already vast distances increase as the search for life on the more distant planets expands.
From a mobile on Mars, it takes 20 minutes for the data to be sent home; from Pluto it is more than four hours.
It is also really expensive to send data across space. “Data from a rover on Mars can cost up to 100,000 times more than data from your cell phone, so we must make those bits as valuable as scientifically possible,” said Eric Lyness, software leader at the Laboratory for NASA planetary environments. Goddard Space Flight Center (GSFC).
That high cost means that scientists can’t run as many experiments or get rovers to analyze as many rock samples as they want.
“By using AI to do an initial analysis of the data after it is collected, but before sending it back to Earth, NASA can optimize what we receive, greatly increasing the scientific value of space missions.” Lyness said.
Lyness and her colleague Victoria Da Poian have trained AI systems to analyze hundreds of rock samples using data from the Mars Organic Molecule Analyzer (MOMA), an instrument within ExoMars Rosalind Franklin Rover, which will land on Mars in 2023.
The mission was originally supposed to launch in a few weeks, but had to be delayed.
The ExoMars rover from the European Space Agency is preparing to leave Airbus at Stevenage. the … [+]
Able to identify organic molecules, MOMA will search for past or present life on the Martian surface and subsoil. The first experiments indicate that the AI has an accuracy of 94%.
“We need to prioritize the volume of data we send to Earth, but we must also ensure that doing so does not yield vital information,” said Da Poian, who with Lyness has been developing smart algorithms.
Given MOMA’s ambitious mission, this AI could prove critical. “The samples will only remain in the scout vehicle for up to a few weeks before the recipient drops the sample and moves to a new location to drill,” said Lyness. “So if we need to re-analyze a sample, we must do it quickly, sometimes within 24 hours.”
Such autonomous systems will be even more important for possible future missions to other potential hot spots for life elsewhere in the Solar System. “As we move forward to explore Jupiter’s moons, like Europe, and Saturn’s, like Enceladus and Titan, we will need to make real-time decisions at the site,” said Lyness. “With these moons, it can take 5 to 7 hours for a signal from Earth to reach the instruments, so this will not be like controlling a drone, with instant response.”
The system will cross-reference a database, essentially giving scientists on Earth a message that says something like:
“I am 91% sure that this Martian rock sample corresponds to a real world sample and I am 87% sure that it is phospholipids, similar to a sample analyzed on July 24, 2018, and this is what those data looked like” .
So all he’s really doing is helping scientists prioritize what’s being investigated further and helping them make decisions more quickly, so they don’t miss out on something really important.
“One of the main challenges for planetary missions is getting data back to Earth: it costs time and money,” said Dr. Joel Davis, a postdoctoral fellow in planetary geology at the Natural History Museum in London, and is not connected to the investigation. . “Given the limited lifespan of the missions, scientists have to be very selective about the data they choose to bring back.”
“These results certainly look promising; Having greater autonomy on board spacecraft is one way to ensure the usefulness of the returned data. ”
Wishing you clear skies and wide eyes.
.
