Artificial Intelligence for Space Exploration

by David Vivancos

Last Friday I was invited to be on The Space Show by David Livingston to talk about “Artificial Intelligence for Space Exploration”.

It was my second time on the program, first was a decade ago, talking about “Nanotechnology for Space Exploration” , when I created my first Space Startup.

David Livingston work is amazing spreading the word on what Space is and what Space should be, and has interviewed many key players in the field, from Astronauts, Top execs from Space Agencies, Private Space Contractors, Space Entrepreneurs, Space Tourism innovators or Private Space enablers like Elon Musk from SpaceX.

So, it is fair to say that David’s job is a true gem of knowledge and Sharing to be recognized with almost 3,000 shows in about 15 years.

In my last show, I talked about, the state of the Art of Artificial Intelligence, and why we are on it’s “golden age”, with a summary of history and key advances in the field, in which I am privileged to be involved for the last 21 years.

Space maybe one the sectors where AI has been more developed since its early days, and computation in general is one of the key enablers for us reaching the Space frontier.

For example, NASA holds conferences about AI, Automation and robotics since the 80s and even the simpler astrodynamics need a great deal of computing resources to be performed.

We talked about AI for Space in 4 areas:
1.- Space Access
2.- Space Structures & Manufacturing
3.- Life Support
4.- Space Exploration

1. Space Access
Reaching Space is one of these elusive endeavors, with a great burden due to the cost to lift a pound or kg there. We must remember that to date less than 500 people have flown there.

It is true that private space access is rising, and will be a reality thanks to Elon Musk, Burt Rutan, Jeff Bezos and others, but we still need to go beyond reusing a rocket to really dropping the cost to lift a kg to space to regular airline figures.

How can we change this? , my first approach 12 years ago, was through the use of Nanotechnology, I even build a startup with that in mind and invest a lot of money and time researching in that space. But the truth is that I haven’t been able to test my research yet, because the environment was, and still is not ready, even with current advances in graphene.

So we need to look at different alternatives. My view is that AI will be key in going from simulation over human premises to simulation over all the spectrum of possibilities, we need to have AI’s that “think” beyond what we humans are capable of.

The new materials research must be conducted by AI if we want to go faster , the rocket equation is what it is, but to find the cheapest combination of materials that could work is something where AI could be key, to speed up things.

At the end what we need to build is basically AI scientists.

We as humans have trouble visualizing and working with more than 3 dimensions, on the contrary AI can deal with millions or billions of them, when I say dimension I am referring to the underlying structure that every problem to solve has.

AI is better that us in this high dimensional space, and the leap ahead seems to even beyond our own comprehension.

2. Space Structures & Manufacturing
Robots are one of the key subjects being developed by Space Agencies since the begging of the Space race, and we have seen incredible progress in this field.

But it is fundamental to build structures directly in space or the surface of any moon, planet or asteroid either from raw or slightly processed materials, and for that we will need autonomous robots to build the structures of the future and mine the raw materials needed.

Even we could start seeing materials or products being built in space and coming back to earth if the process is better performed in the special characteristics there.

So for a matter of cost and matter of Speed, using robots to build and maintain everything out there is without doubt the way to go.

3. Life Support
Even now, with just a few humans living in the International Space Station on a permanent basis, and with the Chinese Tiangong empty at the moment, the problems and challenges we face for living in a very reduced gravity environment are huge. and basically, our body is not built for living there.

Also the current advances in robotic medicine are incredible, the first fully robotic operation in the history of humanity was performed some years ago by two robotic doctors “Da vinci” the surgeon & “Sleepy” the anesthesiologist, it is fair to say that was 100% robotic but not yet 100% autonomous, there was no other person in the operating room, only the patient, but the human doctors were outside performing the operation.

But last year another fully autonomous robot called STAR won a competition with humans in a test set operation with a pig, not yet with a human, but still a progress in that direction.

So it is fair to say that soon we will be capable of performing 100% autonomous and robotic operations. and this AI Robotic Doctor will be key if we want to live and work in space.

We need to have also intelligent space suits, that could adapt to the changes in the environment, like sudden changes in radiation due to solar events, and exo skeletons to work in outside our space habitats and enhance our capabilities.

4. Space Exploration
Here the number of possibilities increase vastly, and it is also one of the areas where artificial intelligence has been more successful and also where the biggest failures have played out.

In fact, Deep Learning and Convolutional Neural Networks are disrupting from the last 3 or 4 years the capabilities of machine vision, way beyond what was possible just 5 years ago.

Image Processing and Object Recognition are now a domain of machines, performing better than humans in many tasks, even with about 80% of our brain devoted to vision and being the best creature so far at pattern recognition.

These techniques are widely used now in almost every field, and it is starting to permeate to space too.

Think of the Big Data from the images captured through all the space proves out there, and by the scientific instruments here on earth, AI will allow us once and for all to understand what we have captured, and the science to come just blows my mind.

Another interesting area is the Next Gen Unmanned space proves, I am all to putting humans in space as soon as possible, but for some scenarios they are and they will be the best technology to use, and here you will need full autonomy even more than what we have in today’s spaceships.

Another interesting area is Swarm Intelligence, with multiple robots working together to perform any task, like mapping a planet, or building a telescope or just exploring and sharing insights.

Also for Ground Based operations further automation and processing of raw data are hot topics in the subject.

From my point of view Artificial Intelligence will play a significant role in the future of Space Exploration and will help us becoming finally a space faring society.