Here’s how AI Is changing NASA’s Mars rover science

Some scientists dream of exploring planets with “good” spacecraft that know precisely what information to search for, the place to seek out it, and the best way to analyze it. Though making that dream a actuality will take time, advances made with NASA’s Perseverance Mars rover supply promising steps in that path.

For nearly three years, the rover mission has been testing a type of synthetic intelligence that seeks out minerals within the Crimson Planet’s rocks. This marks the primary time AI has been used on Mars to make autonomous selections based mostly on real-time evaluation of rock composition.

The software program helps PIXL (Planetary Instrument for X-ray Lithochemistry), a spectrometer developed by NASA’s Jet Propulsion Laboratory in Southern California. By mapping the chemical composition of minerals throughout a rock’s floor, PIXL permits scientists to find out whether or not the rock fashioned in circumstances that would have been supportive of microbial life in Mars’ historical previous.

Known as “adaptive sampling,” the software program autonomously positions the instrument near a rock goal, then appears at PIXL’s scans of the goal to seek out minerals value analyzing extra deeply. It is all finished in actual time, with out the rover speaking to mission controllers again on Earth.

“We use PIXL’s AI to dwelling in on key science,” stated the instrument’s principal investigator, Abigail Allwood of JPL. “With out it, you’d see a touch of one thing attention-grabbing within the information after which must rescan the rock to check it extra. This lets PIXL attain a conclusion with out people analyzing the information.”

Knowledge from Perseverance’s devices, together with PIXL, helps scientists decide when to drill a core of rock and seal it in a titanium metallic tube in order that it, together with different high-priority samples, could possibly be dropped at Earth for additional examine as a part of NASA’s Mars Pattern Return marketing campaign.

Adaptive sampling will not be the one software of AI on Mars. About 2,300 miles (3,700 kilometers) from Perseverance is NASA’s Curiosity, which pioneered a type of AI that enables the rover to autonomously zap rocks with a laser based mostly on their form and shade.

Learning the gasoline that burns off after every laser zap reveals a rock’s chemical composition. Perseverance options this identical capability, in addition to a extra superior type of AI that permits it to navigate with out particular path from Earth. Each rovers nonetheless depend on dozens of engineers and scientists to plan every day’s set of lots of of particular person instructions, however these digital smarts assist each missions get extra finished in much less time.

“The concept behind PIXL’s adaptive sampling is to assist scientists discover the needle inside a haystack of information, releasing up time and power for them to concentrate on different issues,” stated Peter Lawson, who led the implementation of adaptive sampling earlier than retiring from JPL. “Finally, it helps us collect the perfect science extra shortly.”

Utilizing AI to Place PIXL

AI assists PIXL in two methods. First, it positions the instrument good as soon as the instrument is within the neighborhood of a rock goal. Situated on the finish of Perseverance’s robotic arm, the spectrometer sits on six tiny robotic legs, known as a hexapod. PIXL’s digicam repeatedly checks the gap between the instrument and a rock goal to assist with positioning.

Temperature swings on Mars are massive sufficient that Perseverance’s arm will develop or contract a microscopic quantity, which may throw off PIXL’s intention. The hexapod robotically adjusts the instrument to get it exceptionally shut with out coming into contact with the rock.

“Now we have to make changes on the size of micrometers to get the accuracy we want,” Allwood stated. “It will get shut sufficient to the rock to boost the hairs on the again of an engineer’s neck.”

Making a mineral map

As soon as PIXL is in place, one other AI system will get the possibility to shine. PIXL scans a postage-stamp-size space of a rock, firing an X-ray beam 1000’s of occasions to create a grid of microscopic dots. Every dot reveals details about the chemical composition of the minerals current.

Minerals are essential to answering key questions on Mars. Relying on the rock, scientists may be on the hunt for carbonates, which cover clues to how water could have fashioned the rock, or they might be in search of phosphates, which may have supplied vitamins for microbes, if any have been current within the Martian previous.

There isn’t any approach for scientists to know forward of time which of the lots of of X-ray zaps will flip up a specific mineral, however when the instrument finds sure minerals, it may possibly robotically cease to assemble extra information—an motion known as a “lengthy dwell.” Because the system improves via machine studying, the checklist of minerals on which PIXL can focus with an extended dwell is rising.

“PIXL is form of a Swiss military knife in that it may be configured relying on what the scientists are in search of at a given time,” stated JPL’s David Thompson, who helped develop the software program. “Mars is a good place to check out AI since we’ve got common communications every day, giving us an opportunity to make tweaks alongside the way in which.”

When future missions journey deeper into the solar system, they will be out of contact longer than missions at present are on Mars. That is why there may be sturdy curiosity in growing extra autonomy for missions as they rove and conduct science for the good thing about humanity.