Humanity’s Extinction Event Is Coming

Oct 18

Quest for Self-Driving in Space, Miniseries Part 1

I.Overview

Remember the movie Armageddon with Batman Ben Affleck? Or Deep Impact with the magnificent Morgan Freeman? Or maybe The Core with future Two-Face Aaron Eckhart? In each case, we turned an extinction-level event into entertainment because it’s too heavy to think about it happening for real.

But the reality is that an asteroid impact, a change in our magnetic field, or the rising temperature of Earth’s climate are all events that we currently cannot escape. There is no back-up plan. We are, for better or worse, tied to the fate of this planet.

As history has shown, that’s not a good fate to be tied to.

In fact on September 7, 2016 a 30-foot asteroid flew between the Earth and the Moon. Our most powerful instruments only detected it with two days notice. Two days. If the asteroid was only 1000-foot wide, it would destroy all human life and we’d have no back-up to get out of it. Even the White House is worried about it.

Five, yes five, major extinction events have occurred on our planet that we know about. We’re due for another. And when that happens, what’s our alternative? You can’t move to another house. You can’t buy survival, even with a billion dollars in the bank. The only way out, is up.

We must find a way to become multi-planetary if we want to save humanity, your family, and yes, even yourself. Only this can restore the honor we seemed to have lost from the brave days of the 60s, while also ensuring our survival. It’s for the species, folks.

And as a species, we have not allowed ourselves the opportunity to blast off for the stars. Only the space race in the 60s when we were afraid enough of a self-inflicted global extinction event (read: nuclear) that we put forth the funding required to launch into orbit and onto our moon. We didn’t have calculators back then, and now we have supercomputers in our pocket, but no one is allowed out of our atmosphere, save for a few communication and spy satellites.

Doesn’t that make you mad?

It’s not some oppressive government that tells us no. It’s us. We pay our taxes. We elect leaders. Those leaders choose Defense as the primary budget line item, but forget about defending against the forthcoming apocalypse. Funding for NASA in the United States has decreased from 4% of the national budget in the 60s to about 0.5% from 2010 onwards.

That’s just the money side. But in order to move past this threshold from our home planet to space and then onto other planets, we need to do two things:

Travel there.
Survive.

Luckily, we can simplify the problem of this passing this barrier by sending machines in our place. Like TARS from Interstellar, they can go places humans cannot and explore the environment for habitability and resources, even in particularly hostile conditions. Maybe not black hole hostile, but definitely Mars hostile, as the Curiosity Rover has shown.

Only now, with a few bold, private startups are we beginning to see a re-emergence of the space industry. We are about to pass a few very important tests that allow us to explore and visit the cosmos.

The first is launching physical things into space. This is the catalyst that will jump start a new space race. Prices of sending cargo are falling dramatically, down to nearly $500 per pound of payload with SpaceX’s Falcon 9 heavy re-usable rocket. Note that the re-usable part is key. We can’t throw away our “space car” every time we Uber it. And once that becomes standard and cost-optimized we might be able to get that down to $10 per pound.

Imagine what could happen when it costs the same amount to ship something across town as it does into space.

The second, and this is just as important, is the wave of autonomous machines. Tesla has popularized the notion of self-driving cars. SpaceX lands their rocket onto a small barge in the ocean autonomously. Companies are buying startups in the space.

Self-driving will be our gift, our talisman, on the quest to save the species by becoming multi-planetary.

II. Shipping Ourselves to Space

The graph below is from the Founders Fund manifesto, showing the decreasing cost of launching something into space. It begins with the 1960s US-versus-Russia space race and extends to the present day SpaceX-versus-Blue Origin reusable rocket race.

The cheapest method we have today is SpaceX’s Falcon series rockets. With the Falcon 9 Heavy, it’s predicted launching cargo into space will be cheaper than ever before, at $750 per pound of payload delivered to low earth orbit (LOE)on an expendable rocket.

You have to note here, however, that these statistics are as cheap as possible. It costs more to deliver payload on a non-reusable rocket, and on something that’s further out than LEO, like geosynchronous orbit, or to Mars.

For example, based on SpaceX’s published pricing, it would be at least 4x more expensive to deliver far less cargo to Mars.

So what happens when we reduce that cost to $10 per pound? Namely, an explosion of startups, much like iOS. Instead of pushing to production for your continuously deployed web and mobile app, we will see future developers push to production by deploying physical things into space.

“STAGE” takes on an entirely new meaning for software developers when it means your automated regression tests fail, it could blow up a rocket and hurt people on board.

That’s why SpaceX and Blue Origins exist. To make this continuous-deployment-to-space process as cheap and fast as possible. By Elon’s calculations, every 15 minutes.

III. Self-Driving Space Explorers

The most successful products for space, at least in the beginning, will make money by pushing this stuff into orbit. Things like science experiments and new 3D printers.

A company called Made in Space creates a number of these products, including the empty box you see below used for sending things up with Blue Origin.

The box shown in gray is a specialized 3D printer that works in zero gravity. Remember how most 3D printers work. It squeezes out a single layer of liquid ooze, and then another, over and over again until it builds up enough vertically that it creates an object. This can be simple plastic or more esoteroic metals. But when you’re “dripping” something, held down in place by gravity, the entire process has to be re-imagined for space. Things in zero-G would just float away.

Enter these chaps.

There’s also the very real need for oxygen, food, water, and shelter from the harsh elements. Funny how we will end up recreating Maslow’s Heirarchy in every new voyage or planetoid we want to colonize.

And space mining is off to the races with the recent announcement of Deep Space Industry’s Prospector-1:

Their vision is to extract water from asteroids and use the chemical components to hydrate us, but also as oxygen (breathing) and hydrogen (fuel). To do that, you have to identify candidate asteroids, physically get to them, land and attach, and then do surveying, prospecting, and extraction.

In short, you’re going to need some level of self-driving capabilities to make this happen. And wouldn’t it be nice if it “just worked” right out of the box.

Unfortunately, in space you don’t have fleets of these space craft, millions of miles of training data, maps, or an internet connection to the cloud so how the heck are deep learning algorithms going to work?

I don’t think they will. And that’s what I believe we need a better approach.

IV. Enter the Hero of Our Story

There’s a man who, for the last twenty years, has been building artificial neurons. That may sound silly, but remember, the current state of the art for artificial intelligence is just a bunch of math equations linked together, end-to-end.

You send a single piece of data through that equation once and get something out the other side. Do that with enough data points and you’ve got yourself a pretty good approximator. Like how the square footage of a home approximates the price of that home in a certain area. Did you know that’s considered artificial intelligence?

Seems pretty basic doesn’t it?

That’s why, over the last 3 years, this man has dedicated himself to the pursuit of a singular mission for these artificial neuron’s he’s been practicing with for decades. The goal is to make a copy of all the little connections inside an animals brain and nervous system to understand how nature’s millions of years of evolution could create such an efficient mobile moving machine.

Because we are nowhere near the capabilities of mapping the connections inside a human body, he had to start with something much simpler. In this case, the simple C Elegans roundworm. It’s only about a millimeter long and in 1986 it became the first living animal to have its connectome mapped. That’s just a fancy term for the connections and electrical pathways present in a body. Sort of like the electrical wiring in your home.

Even though it’s far from being as advanced as a human, our friendly worm can add Astronaut to its list of accomplishments. It’s survived on the International Space Station as well as the Space Shuttle Columbia disaster back in 2003.

In short, this little wormy earned its little space wings.

So who is this man? His name is Timothy Busbice. He’s one of the co-founders of the OpenWorm project which helped explore the innerworkings of this worm’s connectome. He made a copy of the biology and pasted it into working software. That was back in 2010.

Over the next few parts of this miniseries, we’ll describe how a lone wolf hacking away in his garage could be the researcher who enables our species to save itself on Earth’s roads as well as those on the Moon, Mars, and asteroids.

V. The Future

How do I know this? Because at the beginning of 2016, Tim and I started chatting about the merits and limits of the current approach to artificial general intelligence and what we both thought was a better way.

That little worm brain got plugged into a little robot. And damn if it didn’t avoid obstacles better than all the startups trying to build self-driving cars. Said differently, level 5 autonomous wayfinding. Getting from point A to B without any human assistance.

Bridges ain’t a problem for this little guy. Nor is a lack of internet, training data, maps, or millions of miles of driving. The best part? It automatically comes when you call it, no programming needed.

Here little wormy, wormy!

— Sean