Imagine the same project (Lunar base? Moon telescope? Rocket fuel harvester?) but in the Sahara. Now add the constraint that you want it there, but without ever having any humans on site.

That is what needs addressing. That level of automatisation is the hard part about space.

Indeed. Slightly tangential, but when you start thinking about a Mars colony -- but here on Earth -- a self-sufficient hermetically-sealed environment somewhere underground -- it really starts to drive the scope of the problem home.

We don't really know how to create a self-contained ecosystem here on Earth, with everything on-hand. And the residents won't even all die if someone accidentally pokes a hole in the wall in the Antarctic prototype base.

Indeed. If we're going to be serious about extraterrestrial colonies we should be workshopping our solutions right now here at home where the problems and consequences are orders of magnitude lesser.

There was the Biosphere II project from the 90s, but it ultimately failed.

https://en.wikipedia.org/wiki/Biosphere_2

There was still some very cool engineering that went into making what it was supposed to be. Check out this video on the 'south lung' youtube.com/watch?v=Wind4fjbt_k

Because they didnt want air going in or out of the biosphere they had to figure out a way to account for the massive change in volume as the air temperature inside changes between super hot days and cold nights, otherwise you'd blow the windows out of the thing during the day.

Run by Steve Bannon, of all people!

"Steve Bannon left Biosphere 2 after two years, but his departure was marked by an 'abuse of process' civil lawsuit filed against Space Biosphere Ventures by the former crew members who had broken in.[70] Leading managers of Biosphere 2 from the original founding group stated both abusive behaviour by Bannon and others, and that the bankers’ actual goal was to destroy the experiment."

Shocking ...

From reading the Wikipedia article, it sounds like "run by" overstates his role, but wow nevertheless.

I believe bannon came in after the experiment had already concluded.

I don’t know if I’d call it a total failure. There’s a pretty cool documentary about it an Apple TV+. I think if they had communicated the caveats better and more honestly then it would have been a success on the grounds of being a baby step towards the direction of the more ultimate case. I’d say the biggest success was showing that even with all the resources on earth available, this problem is non-trivial.

Low oxygen, low food, crashed ecosystem. Interesting science, but a total failure at demonstrating the feasibility of a closed ecosystem.

I don't see it as a failure? The first attempt needed a relatively small amount of oxygen injection (which is very much achievable for a Mars mission). The second achieved self-sufficiency.

Really more of a PR spectacle than a worthwhile experiment.

Including humans seems to have caused most of the problems there. If you actually wanted to experiment with closed biological systems in a useful way then I'd recommend doing a bunch of trials with a less cantankerous species of large omnivorous mammal, like the goat.

Biosphere 2 was originally meant to demonstrate the viability of closed ecological systems to support and maintain human life in outer space[0]

>including humans seem to have caused most of the problems

I chuckled a bit at your reply hvd.

The hardest, seemingly intractable problems in the world aren't engineering problems, they are human ones. If we want to level up as a species those are the ones we will need to figure out how to solve.

One approach is to not try to make our colonies self-sufficient. Instead send a study stream of cargo missions to resupply them.

At first, that seems completely impractical due to the enormous fuel costs but it turns out that if you are in this for the long term there is a way around that.

Getting around the solar system is expensive if you are in a hurry, but if you can take your time it can be cheap.

It is not just actual planetary bodies that a spacecraft can orbit. It turns out there are also orbits around Lagrange point. Some stable, some unstable.

If you want to get from planet A to planet B cheaply and aren't in a hurry, what you can do is first put your cargo container in an orbit around an appropriate A/Sun Lagrange point, then nudge it into a carefully chosen unstable orbit.

The unstable orbit gets farther and farther away from the Lagrange point. At some point, it crosses an unstable B/Sun Lagrange orbit, with a low enough delta-V between the two that a nudge can move it from the former to the later.

It then gets closer and closer to that B/Sun Lagrange point, until you reach a point where another nudge can move it into the same orbit as B, just ahead or behind. Another nudge gets it to B.

How long this takes depends on A and B, and it is all over the place. Some combinations take a decade or two. Some take hundreds of years. Some take thousands.

This then is a plan for a long term thinking civilization to colonize their solar system (I'm going to assume that they also call their planet "Earth" and a major moon of it "Moon"). First, get yourself a serious presence in Earth orbit and Moon orbit, and maybe also on the Moon. Anything you can make in one of those places that is needed for your deep space missions is a win because you won't have to get it out of Earth's gravity well.

When you've got to the point that you can produce a steady stream of unmanned space cargo containers loaded with non-perishable items in Earth orbit or Moon orbit, you can start moving them to appropriate Lagrange points, and nudging them into appropriate unstable orbits for the places you want to colonize.

It will take a long time, but when you have a good number of shipments accumulated at a place you want to colonize, you then send your colonists using the fast, expensive route. Note that your colonists only need to take enough supplies for the trip itself. (And they don't need to take return fuel if not everyone is going to stay, as that can be included in the pre-sent cargo).

Here are a few articles on this [1][2][3].

[1] http://www.gg.caltech.edu/~mwl/publications/papers/IPSAndOri...

[2] https://en.wikipedia.org/wiki/Interplanetary_Transport_Netwo...

[3] http://www.dept.aoe.vt.edu/~sdross/papers/AmericanScientist2...

There is a saying that once you reach the orbit (here, on Earth), you are halfway to anywhere.

Lifting objects from Earth's gravity well is expensive. There would be a better case for building extensive industrial base on the Moon, with its 0.16g gravity, and supplying all the distant bases from there.

You can, for example, plausibly build a space elevator on the Moon [1] with contemporary materials (even Kevlar would suffice), thus lowering the price tag enormously.

[1] https://en.wikipedia.org/wiki/Lunar_space_elevator

Doesn't really change anything because you are pretty much out of the Earth's gravity well by the time you get to the moon.

Sure it does — you shuttle humans up to LEO at the current SpaceX Dragon cost, and instead of docking to the ISS they dock to the full-size replica of the Enterprise that an eccentric billionaire 3D printed out of Moon rock and filled with space-grown food, plants, and spandex costumes.

To be fair, we have never had a reason to do so. It is a hard problem, but it isn’t impossible.

Of course we have reason to do so (to test whether we can do it) and projects like Biosphere 2 suggest that while it's not impossible, it is for the time being unsolved.

I meant other than purposes related to space exploration and general scientific research.

I fully support such activities.

Nobody has ever constructed an independent human-compatible biosphere. We have 0 evidence that it is possible at a practical scale. N=1.

Is robotics (longterm) a solution here, and if so, what level of autonomy do we need?

Latency between Earth and the Moon is just enough to make realtime command a bit too rough to use I'd imagine, so I imagine we'd need at least low level autonomous tasks of "move resources from landing pad to storage" and "prepare build site" with enough intelligence to detect when these tasks hit an exception.

On top of that, I suspect we'd need enough robotic automation to be able to build, maintain and run whatever the thing is. That probably means several specialized robots (packing dirt for a foundation, deploying construction pieces, a crane) as well as some general robots for when things do go off script (haul away a broken robot, right a fallen pylon, remove debris).

Does this feel like something achievable in the next decade or two? Are there large pieces we're missing?

Robotics in hard vacuum with abrasive soil, extreme temperature fluctuations, and bullets raining down. We don't deal with that on Earth.

Except for abrasive soil, all the challenges exist on the orbit as well. And various satellites seem to cope.

If attrition of the robots turns out to be a factor, it could still be solved like this: expect that an average robot can survive 6 months on the lunar surface. So send twenty of them to the remote base, build a cave for the reserve robots and "burn" through the reserve slowly. Every 4-5 years, send a human crew to replace or repair the dead robots.

Can be done, the main question is the price tag.

Some of the deep sea robots used on oil and gas rigs come close IMHO.

I smell an XPrize!

Musk is claiming that Starship launches will be closer to $10M than $100M:

https://twitter.com/elonmusk/status/1328770804222468097

(I've heard some ad hominem attacks that this will never happen, but haven't seen any actual evidence that it's impossible.)

This continues to befuddle me.

So let's assume Musk nails it and can do this for $10M. Instead of my 20, imagine 200 100-ton landings of cargo on the moon, say over the period of ten years.

Instead of everybody sitting on Earth, trying to design-session-out the perfect lunar base, why not just keep regularly-delivering supplies, then learn how to build a lunar base while on the moon, figuring it out as you go? After all, just like SpaceX, we're not trying to build a lunar base. We're trying to build a factory for constructing lunar bases, eventually hundreds of them. Solving the modular/generic problem is what we want to do, not build another Space Shuttle.

Perhaps I'm wrong, but it seems like most of the people here, indeed most of the aerospace industry, naturally prefer these big, lumbering, paperwork-heavy, long-lasting bureaucratic programs. But without enough regularly-arriving supplies, why try to solve something up-front, all-at-once, and perhaps years or decades ahead of time when you can just figure out the bare minimum of things you need just as you need them?

But like I said, I'm probably missing something. I've been saying for 20+ years that our real problem is cost-to-LEO. Now that we're just barely beginning to address that, the next problem very well may likely be changing the way we think of large space-based programs. That could very well end up being a bigger problem than cost over the long run.

ADD: Seems to me that the Starship is a solution to delivering humans to space. We need reusable heat shields, booster packs, and payload capsules, along with an Earth-based mass-driver, to truly drive down cargo rates another 100x. At that point we'll begin seriously talking about becoming a space-faring species.

> So let's assume Musk nails it and can do this for $10M. Instead of my 20, imagine 200 100-ton landings of cargo on the moon, say over the period of ten years.

That cost is predicated on rapid relaunch, with multiple launches /per-day/. It wouldn't be ten years, more like two to four :)

The cost of rocket launches is not the largest part of creating a lunar base, designing and manufactoring all the landers, base, electronics, software etc. is the larger part of it.

If I recall correctly, for the apollo program about 30% was for the launch vehicle.

It seems like this would be somewhat of an interrelated valuation.

If launch costs decrease by an order of magnitude, reliability of payload can be relaxed, and therefore costs of payload engineering and manufacture decrease.

If mass reduction is no longer super critical, you can use heavier but cheaper materials that provide the needed strength.

There could be other knock off effect of cheaper and more frequent space launch - you can send multiple versions and see which works the best, ship prototypes back for extensive evaluation, send robots or people to fix broken things.

A lot of things are done differently if you have just a couple super expensive shots per decade & everything needs to work without external help at the first time.

I would assume engineering and building one of these things is non-trivial, and there may also be certain aspects tailored to the specific launch vehicle, so it’s not like as soon as SpaceX proves they can do what they plan to do that you can just pop a telescope on a rocket and go.

Launch providers list the characteristics of their vehicle and the customer has to ensure to stay within the constraints.

SpaceX is special (or used to be special, maybe others have followed suit by now?) in that they make this information public. See here what loads, accelerations or vibrations the payload has to withstand during a launch on a Falcon 9 or Falcon Heavy launch vehicle:

https://www.spacex.com/media/Falcon_Users_Guide_082020.pdf#p...

More like 5+ launches to get it refueled - the payload of Starship is MUCH smaller than its fuel load.

Fully fuel requires at least 8 tanker launches IIRC. Fully fueled Starship weighs 1,200 tons, 1,100 tons of fuel over 8 launches is nearly 150 tons per launch.

> If they don't, you don't pay.

Do you have to put money down like with Tesla Full Self Driving?