AI On Mars: Interview With Tom Soderstrom, Jet Propulsion Laboratory (JPL)

Humanity is placing its mark all through the world, and now on different worlds and planets as properly. For the previous many a long time, people have despatched spacecraft, rovers, and machines to planets close to and much, so it ought to come as little shock that AI is turning into an growing a part of these distant techniques. After all, if it could actually take hours and even days for communication and instructions to be despatched to and from gadgets sitting on different planets, so imbuing these techniques with clever and autonomous capabilities to function within the harsh environments of outer area and on planetary or rocky our bodies makes a number of sense.

At the Amazon re:MARS convention in 2019, Tom Soderstrom, IT Chief Technology and Innovation Officer at NASA Jet Propulsion Laboratory shared his insights into how AI is more and more being utilized by these distant gadgets and a few insights into how the expertise is being utilized for area and planetary exploration. Mr. Soderstrom’s position at NASA is to plan for future missions and strategize primarily based on the place expertise is heading.

“We have created a sort of innovation experience center where we go into the room to experience the future in the current time,” said Mr. Soderstrom. “The purpose of this is to better understand which technologies are ready to be used, while filtering out which are worth pursuing to benefit JPL, NASA, and the enterprise.”

In his speak at re:MARS, Mr. Soderstrom talked about “democratizing MARS”, referring to machine studying, automation, robotics, and area. Democratizing refers to utilizing expertise waves making it obtainable to folks. According to Soderstrom, this implies to “make algorithms and data available for future explorers to stand on the shoulders of giants, not in decades or years, but now.” 

As for AI particularly, the JPL is making use of many facets of the Seven Patterns of AI to the robots, rovers, and probes heading to distant planets. All these remotely-operated techniques are utilizing facets of AI-enhanced laptop imaginative and prescient to present higher insights into the photographs that the techniques seize. They are additionally making use of assorted facets of autonomous systems and management, utilizing machine studying to assist these techniques deal extra adaptively to modifications. Machine learning-powered pattern and anomaly recognition is essential to each space-bound and earth-bound techniques to present higher insights to all the information flowing forwards and backwards between the assorted gadgets and techniques. AI-enabled predictive analytics and predictive maintenance is likewise extremely essential to efficiency of techniques that must function for years or a long time with no capacity for people to service these techniques. 

“Today we use satellites surrounding earth and in deep space, but say we want to know if something is about to go wrong,” states Mr. Soderstrom. “So, we use machine learning and we use the data from previous orbits or tracks and use current data to determine if things are going wrong through machine learning algorithms.” 

The outcomes are then visualized for folks, so everybody stays up to date. “Rather than calling it artificial intelligence, we call it intelligence assistance. Not only have we found the anomalies people have discovered, but also some that people have not yet found.”

Mr. Soderstrom shared one other instance about predictive upkeep. He defined, “When thinking about what NASA and JPL actually does, tracking space craft throughout our solar system, even beyond. To do this, because the earth is rotating, it requires antenna farms that have the ability to connect and communicate with the spacecraft no matter where it is. For instance, with Voyager it takes nearly 40 hours to send a signal there and back, so one antenna would send the signal, and another would receive it.”

Like with any piece of apparatus, antennas require a number of upkeep. So, AI is at present getting used to assist predict if and when upkeep is required, what elements could also be about to go dangerous and desires repaired or changed. According to Soderstrom, this enables them to modify antennas, then take an antenna offline for repairs with out dropping connection. Furthermore, he defined “this can be democratized to trains, cars, planes, etc. as we can learn from their methods as well to improve our machine learning.”

Besides utilizing AI to manage the assorted techniques on earth and in area, AI is getting used to assist present extra perception into the information and pictures that these techniques are capturing. According to Mr Soderstrom, “Let’s begin with right here at home. There are many things happening at home and what we do at JPL and NASA is to look at earth from space. This provides a very different perspective. There are many satellites orbiting earth, taking around 90 minutes to orbit the whole planet. One use case example of AI is detecting forest fires.”

Therefore, because of AI-enabled satellites, imaging techniques are able to detecting smoke clouds, routinely triggering one other satellite tv for pc to zoom in nearer for verification, location and alerting the native fireplace division, in accordance with Soderstrom. This is only one means autonomous satellites work on our behalf. Another instance he shared can be utilizing satellites we have despatched out to assist us examine water and ice patterns, so we will do machine studying on this and let folks work with actual knowledge. 

As for the way forward for AI, Mr. Soderstrom may be very optimistic. “I believe we have now had a number of AI winters, however we’re due for an AI summer time. I believe there are a number of good issues coming. If we have a look at the six expertise waves coming, one is utilized AI, which inserts the theme of democratizing MARS. Applied AI, amongst different waves, all mix to what we name embedded AI all over the place. Essentially, this is able to end in a private assistant regardless of the place you’re. This may very well be from a smartphone, good watch, or different product.”

Indeed, the hunt for area exploration has resulted in lots of merchandise we use as we speak akin to Kevlar, Velcro, child method, and even security glass. As we proceed to push ahead with our quest to discover area, AI will play an growing position on this and assist each in area and again right here on Earth as properly.


Please enter your comment!
Please enter your name here