Mars Rover Mission Will Use AI to Help Scientists Sort Out Findings

An A.I. system based on Earth will automatically analyze and prioritize data coming back from rock samples that the rover collects to determine what data is worth paying attention to.

An artist’s rendering of the Rosalind Franklin Rover that will be sent to Mars in 2022 for a mission led by the European and Russian space agencies. (ESA/ATG Medialab)

(CN) — A planned space probe mission to Mars in 2022 will use artificial intelligence to help scientists prioritize their time and attention as they try to figure out if there has ever been life on the planet, researchers at NASA said Thursday.

While the space agency already uses A.I. to help its Curiosity rover navigate around Mars, the researchers described the way the technology will be used in the upcoming mission as a significant step toward an ultimate goal of having more autonomous probes explore farther reaches of the solar system.

Reaching that goal could be crucial for the future of the nation’s space program. Without automating much of the mechanics, the missions could become too expensive to justify.

“I do believe that these new missions won’t be possible if we don’t use machine learning or artificial intelligence,” Victoria Da Poian, an engineer and researcher at NASA’s Goddard Space Flight Center, said in an interview.

Da Poian, a self-described “space geek,” has been helping develop the A.I. that will be used on the 2022 ExoMars mission, a project led by the European and Russian space agencies. The mission’s Rosalind Franklin Rover was set to launch this year, but the agencies in March decided to delay the launch so that engineers could make sure the rover was fully prepared to fly.

During the mission, an A.I. system based on Earth will automatically analyze and prioritize data coming back from rock samples that the rover collects. The system will help scientists sort out what data is worth paying attention to and what data is boring background noise from things like the rover’s calibration readings, Da Poian said.

Da Poian presented the first results from a series of tests using the technology during a virtual version of the Goldschmidt Geochemistry conference, an annual science gathering that was set to be held in Hawaii this year before moving online because of the coronavirus pandemic. Eric Lyness, another NASA engineer, has also worked on the project.

The tests showed promising results, the researchers said. The A.I. was able to analyze samples from an unknown compound and categorize it with up to 94% accuracy. The system also correctly matched new samples with previous samples 87% of the time, they said.

The team plans to further “refine” the system before incorporating it into the 2022 Mars mission.

The desire for space probes that are more autonomous in how they explore and study distant planets has a lot to do with efficiency.

NASA plans to launch a probe called Dragonfly to an icy moon of Saturn in 2026. The Titan moon is similar to a very young Earth, the space agency says, and could give scientists a better understanding of how life developed here at home.

Because of the distance, communications between Earth and the probe will be slow, but the science can’t necessarily afford to wait.

“In the future, as we move to explore the moons of Jupiter such as Europa, and of Saturn such as Enceladus and Titan, we will need real-time decisions to be made onsite,” Lyness said in a statement about the team’s research.

“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 an instant response,” he said. “We need to give the instruments the autonomy to make rapid decisions to reach our science goals on our behalf.”

Using A.I. to sort out data from the upcoming Mars mission will be mostly a proof of concept, Da Poian said, as the time demands for that particular mission won’t be that intense. It only takes about 20 minutes to get data back from Mars.

But developing the technology for use in deeper space missions could be essential to the space program’s future. Time is money, as they say.

The basic process of gathering and analyzing data from a distant probe can get expensive, so if an A.I. system is able to know when a particular data set is not worth sending back and can be disregarded, that’s a potential cost saver.

Still, also because of how expensive these missions are, Da Poian said researchers have to be fully confident that the technology will work before deploying it. She noted that NASA’s planned mission to Titan in 2026 is a billion-dollar project.

“We still have six years to be able to develop all these tools to be able to put this on board,” she said. “If we’re not 100% sure the intelligence systems are not going to send back the science, we’re not going to take any risks.”

Da Poian stressed that the goal with the technology is not to replace the human role in space exploration.

“We always want scientists to be in the loop, because machines will never be better than scientists,” she said. “It’s more all those small tasks that can be done by automation, we don’t want the scientists to waste time. We want scientists to make decisions when it will be a science question to answer.”

%d bloggers like this: