The aviators might say - a good landing is one you walk away from.
How exactly do you define a successful landing? What if you're upright but all four legs collapsed and the spacecraft crushed your primary science payload?
There is this experiment in OpenAI Gym to land a spacecraft via reinforcement learning. Is reinforcement learning actually working for this, and are such models deployed to real rockets, instead of "hardcoded" math?
Reinforcement learning does work for this, but it's brittle. You almost always end up with a strategy that exploits inaccuracies in the physics simulation – from the perspective of the RL algorithm, useful quirks of the laws of physics – and so doesn't transfer to reality very well, if at all. Its behaviour outside the conditions observed in training is not guaranteed (or even expected) to be sensible, and even its behaviour within the training conditions is often hard to characterise.
An algorithm designed by people who know what they're doing is usually better. More effort, yes, but rockets are a lot of effort! We can afford to pay the cost for a reliable landing system.
If you go fast enough sideways, the lander goes off screen and you can fall past the ground. Then you can flip for unlimited reward. Unfortunately, you can't then crash to see a final score!
Maybe before sending lots of landers we should build a moon GPS system for guiding to a precise safe landing spot?
But it also occurs to me and therefore probably the actual engineers they should be throwing these landers out of an airplane over a rough, rocky earth desert many times before trying the moon?
If they could get the budget to build a few extra multi million dollar spacecraft and throw them in the sand they absolutely would. Unfortunately all management gave us were these cardboard boxes with a lander drawn on them in sharpie so we'll have to use our imagination.
Celestial navigation only gives you attitude information while in space. The video is talking about aligning the guidance systems with the celestial coordinate frame. The stars are far enough away that it is not possible to get absolute position/velocity information using the stars (unless you use some near-future tech like XNAV, https://en.wikipedia.org/wiki/Pulsar-based_navigation). Of course, the position of the planets can give you general idea about what "side" of the solar system you are on. A star like Polaris can give you a rough idea of latitude, but that is not nearly enough precision to perform a planetary landing.
The Surveyor probes used radar ranging to get relative altitude/velocity measurements. They were not very precise in where exactly they landed. The Apollo landers were semi-manually flown visually in the terminal guidance phase which probably helped with their accuracy.
The vision-based systems used today are much more capable of doing precision landings autonomously once you are close to the surface.
I think the stars, in conjunction with the planets, can give you a half decent fix.
The Soviets started landing on the moon in the 1950s. It wasn't pretty, it was primitive, but it has been done. We (as a species) have since landed on asteroids and planets of sulphuric acid.
Space is hard, but we're literally trying to replicate what has already been done.
> I think the stars, in conjunction with the planets, can give you a half decent fix.
What was lacking then was precision. The landers from the 60s (Soviet and US) targeted landing areas that were massive. You basically ensured that the initial descent trajectory would intersect a certain area (using ground-based orbit determination) and the terminal landing sequence just ensured a soft-landing ... wherever that might be.
What is even more impressive is that they did all that without any digital computer on board. For example, in the Surveyor lander, the landing guidance was made up of analog electronics, using the radar signals to command a thrust-vector and throttle to the gimbaled rocket engines. Here is an interesting read about NASA's surveyor probe, if you want to know more: https://www.sciencedirect.com/science/article/pii/S147466701...
In contrast, the new commercial landers are trying to land at very precise locations within hundreds, if not tens of meters of a certain point. And they are doing that with a fraction of the budget of the programs from the 60s. For example, the entire Surveyor program (with 7 landers) cost ~$469 million in 1966. That is nearly 4.5 billion in 2024 dollars with a large part of that going into R&D. The IM-1 lander in contrast, was awarded $118 million. The Japanese SLIM lander cost $121.5 million.
So for what they have, they are doing really well!
Keeping satellites in orbit around the moon is very expensive. The moons gravity is lumpy, and if you try to orbit far enough that it balances out, then perturbations from the earth and sun still end up destabilizing it. This means it takes a lot of station keeping (firing thrusters, using limited propellant) to maintain an orbit.
I think there are plans for better relays situated around the moon, but a lunar gps system is not likely given the costs and engineering difficulties.
It seems both of these landers fell over due to instrument failures. So o solution that relies on building different instruments is not obviously the right one.
I saw a diagram of a lander, maybe just a prototype, that was constructed like a truncated tetrahedron. It would end up on either the bottom, or one of three sides. Once it landed the three sides would open up. no matter which orientation, that would push it to an upright position.
It was because of weight constraints. Spirit and Opportunity weighed 185 kg, while Curiosity and Preseverance both weighed around 1000kg. At that weight airbags large enough to cushion down the payload sufficiently would weigh too much. Just look at how large the spirit airbags were to support a rover that was only 1.5m square.
That's how the earlier Mars rovers arrived, totally valid concept. But it's expensive to get mass to the Moon, and if you think you can likely land upright with thrusters, why bother with something more complex/massive?
Dynetics’ proposal was the most expensive, and they also couldn’t figure out a way to make it work at the time when the choice was made[1]:
> Of particular concern is the significant weakness within Dynetics’ proposal under Technical Area of Focus 1, Technical Design Concept, due to the SEP’s finding that Dynetics’ current mass estimate for its DAE far exceeds its current mass allocation; plainly stated, Dynetics’ proposal evidences a substantial negative mass allocation. This negative value, as opposed to positive reserves that could protect against mass increases at this phase of Dynetics’ development cycle, is disconcerting insofar as it calls into question the feasibility of Dynetics’ mission architecture and its ability to successfully close its mission as proposed.
You can read NASA’s full thoughts on that link. But the basics are, they thought there were good ideas, but they weren’t comfortable picking a lander that went far above the allotted budget, while the team who made the lander wasn’t able to come up with a way that it would work yet.
IIRC the original Dynetics proposal had a positive mass allocation, but it relied on refueling at the Artemis Lunar Gateway. NASA then changed the requirements so that the Artemis Lunar Gateway would not be fully available for the first moon landing, and Dynetics was unable to adapt their proposal.
they could have tossed something at least, somehow they managed to do all that footage without anything I can use to combat the conspiracy theorists who just say the footage is slowed down
In this vein I've always thought it would be cool to see what kind of sports we could have in a zero gravity stadium in orbit. Something like the training facilities in Ender's game.
Maybe a sport like Quidditch could become a reality
Yeah, I can't remember the show, but it was some time travel thing, and one of the tech guys isn't as excited as everyone else at the prospect of visiting great moments in history. They ask him why, and he says, "well, I'm black, so...."
There's probably several different shows that match this description, but you're probably thinking of Star Trek DS9, Season 6, Episode 13, "Far Beyond the Stars".
I wouldn't want to give up my toys. I don't have enough brain cells left to accommodate to 1960s style programming, let alone the outward appearance needed to convince someone to let me use their mainframe back then.
When I was a kid I used my bike to flawlessly get around everywhere I wanted to go.
At some point I wanted to go to places or in an amount of time or with cargo that a bike wouldn't allow.
The Artemis program is doing something similar but not the same as Apollo. It's different so it requires a change in the architecture which in turn means a series of problems because change is hard. Which truthfully is the exact same thing Apollo went through. Apollo was only six for six if you ignore everything that went before it. If you look the events before Apollo 11 it's easy to see that there were difficulties.
That sideways drift was the bane of my existence. I could never quite kill all of my horizontal velocity, and tipping over was common. (Just as the article said)
I wonder how SpaceX will solve this reliably for Starship?
Some Moon ideas:
1) have a small robot bulldozer flatten a landing pad/polygon so there is loads of safe landing area.
2) use same dozer to make a moon highway to other sites of interest around the moon.
3) Moon GPS or laser-based local navigation beacons, so the spacecraft can rely on those if instruments fail.
4) Just stupid-wide landing legs that let the craft ski around the moon even when landing at an angle and with horizontal velocity.
The reason Odysseus's legs are so narrow is that they wanted fixed legs, so there's no leg deployment that could go wrong. But payload fairings are only so wide. The legs are the widest that fit a Falcon 9 payload fairing.
I really like the idea of a mission to build infrastructure though.
Just like Canadarm ([1]) on ISS considerably simplified docking, I can see a similar approach working for lunar landing: slow down to below 5 m/s, get caught by a robotic arm and gently put into a good spot.
A launch and return to the same rotating body is surely a simpler maneuver than orbital transfer to a different spinning body and landing. No doubt we are better at it but I’m not sure earth landings are an accurate analog.
That depends on how much atmosphere. Mars has enough atmosphere to be annoying, but not really enough to be useful. Winds can throw you off target, but you can't use a parachute for a soft landing.
Apollo 11/Eagle was deliberately piloted away from the initial landing site because the pilot did not think the original location was safe. So until a perfectly placed target for a robot to land on like SpaceX does (which seems only likely if establishing a sort of base), the robots will need to be made smarter about landing or built to be more agile on landing on less than ideal sites.
The article mentions that SpaceX has landed on the moon once, and tipped over. The moon is harder to land on then the Earth - less gravity and atmosphere.
SpaceX launched Odysseus. If you want to argue that this doesn't count as SpaceX landing, that's fine, but then it's even less evidence of Spacex's ability to land anything on the moon standing up.
Fair enough; Apollo 12 landed within walking distance of the Surveyor probe, but I don't have enough information to know whether 535 feet from the probe was exactly on target, or hundreds of feet off target. So maybe we could have landed on a pad 50 years ago, too?
... and then Apollo 12 targeted one of the earlier Surveyor probes (which had successfully done its own automated moon landing in the mid-1960s!), and the guidance system had it coming down very close to the earlier probe -- astronaut Pete Conrad manually overrode it to make sure they stayed a safe distance away.
There had been upgrades to software and procedures since Apollo 11 -- most notably, a new guidance parameter the astronauts could enter ("noun 69") to correct for deviations between the planned lunar orbit and the one in which the spacecraft actually was, before descent.
The Autonomous Site Preparation: Excavation, Compaction, and Testing (ASPECT) Project will develop tools and methods to clear, level, and compact the lunar surface. ASPECT is a fully autonomous rover with equipment for regolith excavation, boulder moving, and surface compaction.
There are a number of well fleshed out plans for doing essentially this. One of the more interesting ones for me was that microwaves can be used to fuse lunar regolith into a solid. One plan had a solar powered rover that sat there and microwaved the ground until is was solid far enough down, then would move forward on to that patch and start with the next patch. As I recall it would take about a month to make an acre sized "pad" which would support landing.
You know how the martian landers now use a crane to lower the lander to the ground so the retro rockets are outside of the ground effects range?
Someone proposed doing something similar with a lander, where a landing-site-prep robot gets dropped first. But that sounded like a lot of hovering to me, and a solution that would be very specific to lunar landing. I can't see that being successful on anything with higher gravity.
Better perhaps to send a separate stage or a scouting mission to do the work, then land later.
isn't regolith microscopic and sharp? A sealed ball that rolls around is one thing, but wouldn't a non-sealed ball just get its motor shredded pretty quickly?
I remember reading before that given the consistency and weight of moon dust that electrostatic protection of assets is viable.
I guess the concept would be to surround the motors in a electrostatic force that results in propulsive removal of particulates away from the drivetrain.
You can. But landing isn't the only issue. You have lots of things you need to take into account.
For example, at some point you need to be in the right position for your communication and power needs. The payloads need to handle rolling around. Lots of reasons.
Its a very invasive solution for one particular problem but doesn't help and makes other things harder. If you don't slow down, being round want help you when splatting on the ground.
At the end of the day, we can make landing of the moon reliable with other methods that are less invasive and thus less costly. The solution you suggest is one that you use if you don't have any good options left.
Worth noting is that because Starship HLS carries astronauts, it has to be capable of abort-to-orbit -- that is, to cancel the landing at any point and return to Lunar orbit. The Apollo LEM would have done this by shutting down and dumping the descent stage then lighting the ascent motor: Starship is a single stage that should have enough fuel and oxidizer left after a successful landing to lift off and return to orbit with a minimal payload.
I expect if astronauts aboard HLS lose their altimeter they'd have to abort the landing immediately -- to proceed without it would be the height of recklessness. But Odysseus had no abort-to-orbit capability so was committed to landing.
>I have never once read about abort-to-orbit capability as a concept
ATO was an abort mode [1] on the Shuttle program and is notably the only abort mode that was successfully used in the entire program, on STS-51f [2] . Challenger suffered an engine anomaly on liftoff that resulted in a lower orbit than was intended, but otherwise the mission went off without a hitch.
Thanks! I’d seen about the bailout capability mentioned in passing, but had always wondered what it would be in practice (spoiler: a pole!). Also, I didn’t realize a second engine almost shut down on STS-51-F .
Per the links:
“A particularly significant enhancement was bailout capability. Unlike the ejection seat in a fighter plane, the shuttle had an inflight crew escape system[12] (ICES). The vehicle was put in a stable glide on autopilot, the hatch was blown, and the crew slid out a pole to clear the orbiter's left wing. They would then parachute to earth or the sea. […] Before the Challenger disaster, this almost happened on STS-51-F, when a single SSME failed at about T+345 seconds. […] A second SSME almost failed because of a spurious temperature reading; however, the engine shutdown was inhibited by a quick-thinking flight controller. If the second SSME had failed within about 69 seconds of the first, there would have been insufficient energy to cross the Atlantic. Without bailout capability, the entire crew would have been killed.“
It's doubly confusing that STS-51-F, with the Challenger, is the only exercised launch abort; while STS-51-L is the famous launch disaster for which Challenger is most well known.
> “As a result of the changes in systems, flights under different numbering systems could have the same number with one having a letter appended, e.g. flight STS-51 (a mission carried out by Discovery in 1993) was many years after STS-51-A (Discovery's second flight in 1984).[6] It wasn't until STS-127 in 2009 where the flight numbering system returned to a standard and consistent order.”
Ouch, shortly after they get standardized and consistent flight numbers, the shuttle program gets cancelled. I guess computer science doesn’t have a monopoly over the difficulty of naming things.
Apollo Lunar Module had an abort-to-orbit that was also used to lift off the surface of the Moon after successfully completing the mission. It used explosive charges to throw the lander frame away and involved Apollo Guidance Computer manuevering into orbit at any point of the mission up until the landing.
Dragon 2 has abort to orbit capabilities, too. The abort zones they call out as the rocket's IIP advances up the east coast continue until Ireland, and then after that, it's abort to orbit, where the superdracos will carry the ship to orbit without the second stage.
Abort-to-orbit is a confusing term since it suggests the Shuttle's specific ATO mode. I presume the requirement is "safe abort at all points during lunar descent/landing" rather than specifically to orbit (e.g. an abort mode that put them directly on a return-to-Earth trajectory would probably also be fine).
Sure. I meant more generally, presumably NASA's requirement is "has an acceptable plan to safely return them to Earth after abort" rather than specifying particular orbits.
Huh, then you're one of today's lucky ten thousand!
Apollo 14 had a piece of loose solder in the button triggering abort-to-orbit, so it occassionally triggered itself. This wasn't a problem en route to the moon, but the second the descent phase started it would have been a Poisson-timed bomb that would prevent the landing.
There was a bit of memory that could be set to ignore the state of the abort button (this bit was the reason the abort sequence wasn't triggered en route). The problem was this ignore bit was reset by the landing sequence (to allow aborting once landing started), and they did not believe the astronauts would be quick enough to set the bit again before the button shorted out and triggered the abort.
(Ignoring the abort button was fine because an abort could be triggered in the computer instead. Takes a little longer but was determined a better option than scrapping the mission.)
Don Eyles came up with a clever hack. Setting the program state to 71 ("abort in progress") happened to both allow descent to start and prevented the abort button from being effective. So this program state was keyed in just before descent.
The drawback was that it obviously put the computer in an invalid state so some things were not scheduled correctly but Eyles and colleages had figured out which things and the astronauts could start those processes manually.
Then once the computer was in a reasonable state again the ignore abort bit could be set and the program mode set correctly and it was as if nothing had happened.
Among last-ditch options considered for the Apollo programme (specifically several planned but eliminated long-duration, two-week missions), was the ultimate LESS-is-more approach: "Lunar escape systems".
This was basically a lawn-chair rocket for two which would utilise a disabled LEM's (lunar excursion module) fuel tanks, and would be hand-piloted without any guidance computer to an intercept orbit with the Apollo Command Module, with the hope that a rendezvous and crew transfer could occur within the four-hour window of space-suit oxygen supplies. Given that the CM's orbital period was two hours, this meant at best two chances for a successful intercept.
(I'd run across this from the recently submitted MOOSE article, "Man out of space, easiest", a strap-a-foam-mattress-to-your-ass reentry concept: <https://en.wikipedia.org/wiki/MOOSE>.)
I think since then Elon has mentioned that abort via the main Starship engines may be possible through all points of a launch (putting aside the landing process for now). Probably also helped by the hot staging related changes, since IIRC the concern regarding abort modes was whether or not the engines could safely ignite and separate from the booster.
It does still leave the system without a means of aborting if the ship's main engines have trouble, although I suppose they do have a good bit of redundancy there.
I think GP was saying that the lander is single-stage. By the time of a presumptive lunar landing, there's no lower stage to drop as with the Apollo LEM.
Assuming the landing is soft enough to survive, shouldn't it be fairly trivial for the astronauts to disembark and right the lander? To the extent anything in space is trivial.
Even if it's a bit more than doable by hand a ratchet jack should make short work of it.
Yup, jack points in the right place, jacking equipment with sufficient range of motion, AND solid lunar soil in the right place under the jack points.
Plus, you've got to get the whole jacking operation done without damaging any of your main or control thruster rockets, and without tipping past the upright point over to the other side, or just effectively rolling onto an adjacent side.
I wouldn't want to go on a craft where [jacking it back upright] was in the top ten on the list of recovery options to get home.
>I wouldn't want to go on a craft where [jacking it back upright] was in the top ten on the list of recovery options to get home.
For a counter, I wouldn't want to be on a lander so fragile[0] that being manipulated upright is infeasible. Something will go wrong, maybe not on that lander, but when it does go wrong it'll have to get duct taped together.
[0]Not just mechanically, but in terms of operation scope. Planning that everything must go perfectly or people die is a recipe for the latter.
agree - which is probably why Elon Musk is so obsessive about increasing the efficiency and thrust of the merlin and raptor rocket engines, a huge amount of downstream capability can be achieved by increasing that number (all other things being equal)
I deliberately did not say I wanted jacking to be infeasible — but I DO want it quite far down on the options list, as in there's >10 better things to try first.
(And yes, I've done a fair amount of wrangling vehicles, gear, etc. in snow, dirt, mud, & rocks, and eventually it can often be gotten out. But on a different planet/moon, it really should be not be anything close to a primary option. OTOH, if it's got a set of 6+ pop-out lever-legs to upright itself, tested, etc., that's a different solution)
So if the lander falls over because two of the feet land on regolith, odds are good you don't have a solid point on that side of the craft to put the jack...
All the extra mass budget would likely be better spent on a more robust attitude control system to avoid flopping over in the first place. Unless you need it for something else anyway.
The Apollo LEM, only craft that has ever taken humans to the moon's surface, weighted somewhere around 20,000 kg after landing. Since it only ever operated in space and lunar gravity, it could be built with a much higher mass fraction than a rocket launched from earth requires - greater than 30%, where a Falcon 9 in comparison has a mass fraction below 5%. Even then, the LEM structure had to be built incredibly lightly. While the LEM structure could obviously be lifted by crane and survive launch and docking stresses, those were are at designed points in the structure. Without the presence of a crane capable of lifting the whole LEM, righting an LEM that had landed on its side would have been effectively impossible. Basically all of the modern proposed manned lunar landers are considerably larger than the LEM, and thus considerably heavier.
For comparison, a craft built for earth launch mass fractions probably wouldn't survive falling over in the first place - when that happened to a Falcon 9, the whole rocket simply exploded.
>shouldn't it be fairly trivial for the astronauts to disembark and right the lander?
I don't think that is an assumption you can make. In the worst case scenario the lander lands on the door. In which case the only way to disembark is to lift the lander.
I expect if astronauts aboard HLS lose their altimeter they'd have to abort the landing immediately -- to proceed without it would be the height of recklessness.
In addition to the obvious, it should be taken into account that the absence of atmosphere makes very difficult to assess distance and scale. Videos of approach seem like a fractal browser.
Astronauts must be nervous stepping onto the first manned flight of a new craft that has a 100% success rate in the sole previous flight, but might have only a 50% success rate by the end of their mission...
Well at least they would have no one else to blame but themselves. As far as I remember it was the pilots that insisted on the shuttle to be not be 100% automated, so they had to do it this way. The soviets just made the whole shuttle automated so it could be tested without risk to crew.
With both Crew Dragon and Starship, there will have been _many_ successful missions involving un-crewed variants of the spacecraft (Falcon 9 and Cargo Dragon were both well-proven systems before crew was a possibility).
I couldn't find the relevant XKCD but I remember one along the lines of "why did they just do Y?". I say all of this in jest, as this is what we do here at HN: give our opinion on areas outside our expertise.
A spacecraft perhaps could inflate lots of impact balloons that would cushion impact, allowing landing in any orientation. Then on landing, rotate with gyros or something until the legs are underneath and that way end upright?
Humans did it manually in the 60's. How hard can it be to do it with computers and radar? I don't want to sound like that guy but... how hard can it be?
Edit: My point being: spacex is already doing it on earth, dealing with stronger gravity and air non linearity.
I highly recommend taking a look at Kerbal Space Program plus k/OS. I've done a few Kerbal hackathons with friends where we all get the same spacecraft, in the same scenario, and have to write k/OS scripts to perform some mission. Difficult, but fun and often hilarious!
More seriously, the humans in question were very skilled pilots with huge amounts of general flight experience and specific lunar training; they also had access to hardware that had already been expensively tested in the lunar environment. Neither of those were available to this project.
The IM-1 Lander was supposed to land using a LIDAR altimeter, but they forgot to remove the safety before launch. They tried to make a last minute software change to use an experimental navigation system from NASA to get altimetry, but this didn't work. So the lander landed using visual navigation and IMU data only for the last 15km to the surface.
It probably would have landed upright if the LIDAR worked. It is impressive that it landed as intact as it did
Because streets are not controlled environments. Planes have auto pilots for a long time, because air is a highly controlled environment with professionals agreeing to cooperate and making logical decisions (most of times).
What you do is descend very slowly, as you descend your thrusters will slowly melt a smooth landing pad beneath you. The only problem is you have to sacrifice part of the landing legs when you take off because obviously the melted cheese will bind to them once you touch down.
I'm all for the US space program returning to the Moon and the NASA public-private approach seems fairly reasonable.
It is very odd however to see no mention of the fact that China is about a decade ahead, having completed the Chang'e 3, 4 and 5 missions. which included a successful return of Moon rocks to the Earth. Two more missions are in the works.
The article mentions a Japanese effort, but the omission of China's successes seems fairly deliberate. I wonder what the motivation was?
More power to you, but I've had multiple munar landing attempts turn into munar rescue missions precisely because the first mission lander tipped over :D
Maybe for a small lander, an alternative to landing upright would be to avoid having an upright in the first place. Have a spherical/polyhedral design with legs on all faces and redundant solar panels/cameras on each face.
Mars Pathfinder did exactly this. It was a tetrahedron which could land in any orientation. As it happened they rolled a 4, but if it was on one of the other 3 sides it would have pushed itself upright as the sides folded open.
Very easy solution until you realize you need solar panels facing the sun[1], radiators facing deep space[2], antennas facing earth, payloads facing x, y, and z, deployable payloads facing x, y, and z plus the lunar surface. And all of those things are fragile, at least in terms of mounting them on the outside of a bowling ball rolling across sharp gravel/sand. And of course multiply difficulty by the number of all configurations^2 for both design and testing. [1] and [2] are especially huge drivers, you need power and therefore heat input, but you also need radiators to expel heat. Depending on complexity, solar panels in shade act as passive radiators leaking heat at best, and at worst as heaters, actively taking battery power to heat up and become even better radiators. And radiators act, to some extent, less effectively or even as heaters if they are facing the sun, sunlit earth, or sunlit lunar surface instead of deep space.
It's not only a matter of adding complexity, mass, money, etc. This kind of solution is simple on its face, but it is akin to being ignorant of why a car has windows (for the driver to see), doors (for the passengers to enter/exit) or a cooling system (to not overheat), and then reading that Jeeps (or to be more fair, a new Jeep prototype) are prone to tipping over on a trail after driving 1000km on a highway to a trailhead, and then commenting on how Chrysler should just make a dodecahedron Jeep with 20 wheels instead of 4.
These spacecraft also made it past 99% of the hurdles, the next try will probably make it and overcome the final 1% without starting from scratch and adding all kinds of complexity that are far more likely to add failure.
I work in the industry and opinions are mine and not of my employer.
I think the economics of a one-off lunar lander are a bit different to those of Jeep! The cost of the lander hardware is only a small part of overall cost of the mission, and adding some (or even a lot) of redundancy would seem a small price to pay if that provides a high likelihood of mission success. Of course this needs to be compared to cost and reliability of alternate solutions. This is a lot different to adding a lot of redundancy and cost and compromised design to a mass-produced vehicle where you can iterate and test in deployed environment as many times as you like.
Obviously N-way redundancy isn't optimal, but it would seem potentially simple if you had a modular design with multiple instances of the same "omniface" component. Perhaps use 1/2 of each redundant face for radiator and 1/2 for solar panel?
I'm not suggesting a rolling design - just one with legs on every side such that if it did tip over (or worst case tumble if landing on an incline) from preferred engine-down orientation then it would not make any difference.
Seeing as Intuitive Machines wanted to avoid deployable (spreadable) legs as a way to achieve a more stable low center of gravity, another alternative would seem to be to shrink the design (lose the height) to better work within the width of the faring, although I don't know how viable that would be given amount of propellant needed for the landing.
Cost of hardware is actually a huge part of overall cost, and the economics are worse, not better when you need to prototype and test things for a one-off mission/vehicle made by a new company vs updating a vehicle that sells hundreds of thousands, and making it by a company that has sold millions. There's virtually no technical risk in designing the next iteration of Jeep.
"Just put 1/2 radiator and 1/2 solar" and "just put legs on every side" is still really simplistic and nobody even mentioned thermal insulation yet, which is normally a much higher proportion of surface area compared to radiators and solar, at least if all three are on the main spacecraft body. I was being extreme in my last comment to make a point, but you still sound to me like a guy suggesting 20, and now 12 wheels on a Jeep instead of 4.
> you still sound to me like a guy suggesting 20, and now 12 wheels on a Jeep instead of 4
No - on a lunar lander, not on a jeep. It makes a difference (I'm really not sure why you started talking about Jeeps).
The difference is the cost to test a lunar lander (on the moon) vs cost to test a jeep, and the fact that we're only going to build one lunar lander vs a production run of millions for the jeep.
For the Jeep it's cheap enough to iterate and test (on earth), and end up with an optimized design that has 4 wheels and not 12 or 20. Not only is this relatively cheap to do, but there's a huge incentive to do so since we're going to mass produce the Jeep and would therefore like to reduce build cost and maximize profit.
In contrast, the lunar lander is very expensive to "test" since that means a failed mission to the moon. Maybe that mission costs $5M, so if we fail twice before getting it right we've added $10M to the cost. If we can avoid those failed tests by instead adding $1M of redundant hardware that makes it work first time, then we've saved a lot of money. We're only building one lunar lander, so there's no multiplier in front of that $1M we added to the hardware cost.
The first mars lander was in a tetraheadron in an airbag. There was no way for it to land in the wrong orientation because the tetraheadron always unfolded in the correct orientation. And that was almost 30 years ago.
Yes - that was a great design. I've seen some people suggest that it wouldn't work on the moon due to lack of atmosphere and less gravity, but not obvious why that would prevent it.
Of course it's a more complex design in a way given that you deploy this airbag at low altitude - something else to go wrong. It's interesting though that NASA has used that and the even more complex air-crane technique on Mars both successful first attempt. Even though they seem complex I'd assume NASA determined they were the least-complex approaches that had a high likelihood of success.
Maybe the economics of doing this on the moon, and with a cheaper lander, are different - better to have a simpler system with higher chance of failure, and redo the mission if it fails ?
With a lunar lander you can play with the weight distribution. You can make it more dense on one side. This, together with a self-righting shape, could increase the chances of settling in the desired orientation.
While it was settling it would destroy all/most of the external fixtures including antennae, engine nozzles, and sensor housings. Bonus the rocking action during settling could puncture the hull or break a windows on a sharp boulder.
What if airbags inflated around the lander? The external shape of the lander-with-inflated-airbags and the internal weight distribution could make it self-righting, while shielding the external fixtures and the hull.
Just throwing ideas around and role-playing spacecraft designer.
That shape must be patented or copyright protected, because all the Amazon listings are several hundred dollars each. The usual knockoffs are conspicuous in their absence.
Indeed. The Wikipedia article says: "The gömböc, as the first physical example, is less sensitive; yet it has a shape tolerance of 10−3, that is 0.1 mm for a 10 cm size."
Well, it's an example of a convex homogeneous body that does this. It's much easier if these are not requirements, just have a sphere with an off-center mass.
Radical suggestion. Don't make your lander taller than wide. I'm the first to assume that the science nerds have thought of everything and that my uneducated self has nothing to say about it. But then I saw the lander. It looks very tippable.
So then here's my luddite take. Can't they just unfurl wider stabilizers from the legs that increase their footprint? At slow speeds in low gravity it seems like they wouldn't need to be heavy or strong.
In general, when it seems to my uneducated opinion that a bunch of experts have missed something obvious to me, it's almost always the case that the one who has missed something is me.
I feel exactly the same and this is the reason why I would absolutely love to read an expert article titled "Why super-wide Moon landing gear is not a good idea"
Stabilizers are made of matter, and matter has mass. Lifting mass off of the Earth and on to the Moon takes fuel. Fuel is made of matter too, and has mass. Lifting the fuel to lift the fuel to lift the mass is expensive.
So much cheaper just to have an altimeter and no extra stabilizers. Sadly, their altimeter didn’t work. They forgot to turn it on correctly.
Honestly, they’re lucky that it got so close to landing correctly; if it had been going faster the damage would have been even worse than just a crumpled landing leg.
TIL the lunar lander for Artemis is 100 TONS! My goodness! I didn't think it'd be the Mylar balloon of the Apollo LM, but I did not really think it was going to be the entire first stage of Starship. I guess I need to read more.
320 comments
[ 4.6 ms ] story [ 319 ms ] threadSeems like that would save a lot of fuel too if you don't care how you come down, just not too incredibly fast.
(wasn't there a Mars landing like that)
How exactly do you define a successful landing? What if you're upright but all four legs collapsed and the spacecraft crushed your primary science payload?
https://gymnasium.farama.org/environments/box2d/lunar_lander...
An algorithm designed by people who know what they're doing is usually better. More effort, yes, but rockets are a lot of effort! We can afford to pay the cost for a reliable landing system.
Best I could do was a 970 point crash.
But it also occurs to me and therefore probably the actual engineers they should be throwing these landers out of an airplane over a rough, rocky earth desert many times before trying the moon?
https://www.youtube.com/watch?v=hVCNS2jQQ6g
The Surveyor probes used radar ranging to get relative altitude/velocity measurements. They were not very precise in where exactly they landed. The Apollo landers were semi-manually flown visually in the terminal guidance phase which probably helped with their accuracy.
The vision-based systems used today are much more capable of doing precision landings autonomously once you are close to the surface.
The Soviets started landing on the moon in the 1950s. It wasn't pretty, it was primitive, but it has been done. We (as a species) have since landed on asteroids and planets of sulphuric acid.
Space is hard, but we're literally trying to replicate what has already been done.
What was lacking then was precision. The landers from the 60s (Soviet and US) targeted landing areas that were massive. You basically ensured that the initial descent trajectory would intersect a certain area (using ground-based orbit determination) and the terminal landing sequence just ensured a soft-landing ... wherever that might be.
What is even more impressive is that they did all that without any digital computer on board. For example, in the Surveyor lander, the landing guidance was made up of analog electronics, using the radar signals to command a thrust-vector and throttle to the gimbaled rocket engines. Here is an interesting read about NASA's surveyor probe, if you want to know more: https://www.sciencedirect.com/science/article/pii/S147466701...
In contrast, the new commercial landers are trying to land at very precise locations within hundreds, if not tens of meters of a certain point. And they are doing that with a fraction of the budget of the programs from the 60s. For example, the entire Surveyor program (with 7 landers) cost ~$469 million in 1966. That is nearly 4.5 billion in 2024 dollars with a large part of that going into R&D. The IM-1 lander in contrast, was awarded $118 million. The Japanese SLIM lander cost $121.5 million.
So for what they have, they are doing really well!
I think there are plans for better relays situated around the moon, but a lunar gps system is not likely given the costs and engineering difficulties.
https://en.wikipedia.org/wiki/Mars_Exploration_Rover
https://en.m.wikipedia.org/wiki/Sky_crane_(landing_system)#/...
https://en.wikipedia.org/wiki/Luna_9
For that matter, the Space 1999 "Eagle" for the same reason, just good design. https://news.ycombinator.com/item?id=39484015
If a human lander tips over for any reason, that could be disastrous.
> Of particular concern is the significant weakness within Dynetics’ proposal under Technical Area of Focus 1, Technical Design Concept, due to the SEP’s finding that Dynetics’ current mass estimate for its DAE far exceeds its current mass allocation; plainly stated, Dynetics’ proposal evidences a substantial negative mass allocation. This negative value, as opposed to positive reserves that could protect against mass increases at this phase of Dynetics’ development cycle, is disconcerting insofar as it calls into question the feasibility of Dynetics’ mission architecture and its ability to successfully close its mission as proposed.
You can read NASA’s full thoughts on that link. But the basics are, they thought there were good ideas, but they weren’t comfortable picking a lander that went far above the allotted budget, while the team who made the lander wasn’t able to come up with a way that it would work yet.
[1] https://www3.nasa.gov/sites/default/files/atoms/files/option...
the past is a foreign country
The jumps (4 feet) and bounds (15 feet) are pretty good by everyday standards, but like half of the Earth-bound records.
The revenue for this would pay for several Starships.
Maybe a sport like Quidditch could become a reality
It's a nice place to visit, but I wouldn't want to live there.
Seems pretty sweet to me.
I'd probably be fired when they found out I hadn't read anything written since the 70s....
The Soviet Union is even more foreign with 50's tech and no human to aid primitive computers.
Better failure rate than the last 12 months of international moon landings.
At some point I wanted to go to places or in an amount of time or with cargo that a bike wouldn't allow.
The Artemis program is doing something similar but not the same as Apollo. It's different so it requires a change in the architecture which in turn means a series of problems because change is hard. Which truthfully is the exact same thing Apollo went through. Apollo was only six for six if you ignore everything that went before it. If you look the events before Apollo 11 it's easy to see that there were difficulties.
My landers are designed to land sideways.
I wonder how SpaceX will solve this reliably for Starship?
I really like the idea of a mission to build infrastructure though.
1. https://en.wikipedia.org/wiki/Canadarm
But mostly by obstetricians and context-unaware autocorrect software?
https://dictionary.cambridge.org/dictionary/english/birthing
https://dictionary.cambridge.org/dictionary/english/berthing
Apollo 11/Eagle landed something like 4 miles off from their target site, but SpaceX nowadays routinely lands on target.
Landing on the moon is easier. It just looks hard because it's so expensive to get there that you don't get many attempts.
I'm not an expert but I think landing on Mars with no atmosphere would be harder, because you'd need so much more fuel for a controlled descent.
On the moon?
SpaceX has not yet attempted to land anything on the moon.
There had been upgrades to software and procedures since Apollo 11 -- most notably, a new guidance parameter the astronauts could enter ("noun 69") to correct for deviations between the planned lunar orbit and the one in which the spacecraft actually was, before descent.
https://www.forbes.com/sites/davidmindell/2019/11/19/apollo-...
The topic was exactly that: landing pad preparation solutions. Here's a summary slide for one of the winning proposals: https://www.nasa.gov/wp-content/uploads/2022/03/lustr2021_qu...
The Autonomous Site Preparation: Excavation, Compaction, and Testing (ASPECT) Project will develop tools and methods to clear, level, and compact the lunar surface. ASPECT is a fully autonomous rover with equipment for regolith excavation, boulder moving, and surface compaction.
Someone proposed doing something similar with a lander, where a landing-site-prep robot gets dropped first. But that sounded like a lot of hovering to me, and a solution that would be very specific to lunar landing. I can't see that being successful on anything with higher gravity.
Better perhaps to send a separate stage or a scouting mission to do the work, then land later.
https://www.space.com/jaxa-slim-moon-lander-lev-2-ball-robot
There's also the method of dropping a ball of airbags that deflate and detach, so you aren't limited to being a ball.
I assume it's more efficient and versatile (lots of tools on these things) to not be a ball.
You basically want a small truck loaded with equipment and a solar panel on top.
I guess the concept would be to surround the motors in a electrostatic force that results in propulsive removal of particulates away from the drivetrain.
I don't know if it has been practically used.
https://www.sciencedirect.com/science/article/abs/pii/S00945...
For example, at some point you need to be in the right position for your communication and power needs. The payloads need to handle rolling around. Lots of reasons.
Its a very invasive solution for one particular problem but doesn't help and makes other things harder. If you don't slow down, being round want help you when splatting on the ground.
At the end of the day, we can make landing of the moon reliable with other methods that are less invasive and thus less costly. The solution you suggest is one that you use if you don't have any good options left.
https://en.wikipedia.org/wiki/Luna_9
I expect if astronauts aboard HLS lose their altimeter they'd have to abort the landing immediately -- to proceed without it would be the height of recklessness. But Odysseus had no abort-to-orbit capability so was committed to landing.
I have never once read about abort-to-orbit capability as a concept, let alone a requirement for Artemis HLS.
Here’s a 4 year old video detailing past abort systems and why Starship won’t have one: https://m.youtube.com/watch?v=v6lPMFgZU5Q
ATO was an abort mode [1] on the Shuttle program and is notably the only abort mode that was successfully used in the entire program, on STS-51f [2] . Challenger suffered an engine anomaly on liftoff that resulted in a lower orbit than was intended, but otherwise the mission went off without a hitch.
[1] https://en.wikipedia.org/wiki/Space_Shuttle_abort_modes#Abor... [2] https://en.wikipedia.org/wiki/STS-51-F
Per the links: “A particularly significant enhancement was bailout capability. Unlike the ejection seat in a fighter plane, the shuttle had an inflight crew escape system[12] (ICES). The vehicle was put in a stable glide on autopilot, the hatch was blown, and the crew slid out a pole to clear the orbiter's left wing. They would then parachute to earth or the sea. […] Before the Challenger disaster, this almost happened on STS-51-F, when a single SSME failed at about T+345 seconds. […] A second SSME almost failed because of a spurious temperature reading; however, the engine shutdown was inhibited by a quick-thinking flight controller. If the second SSME had failed within about 69 seconds of the first, there would have been insufficient energy to cross the Atlantic. Without bailout capability, the entire crew would have been killed.“
Why is the naming scheme of shuttle launches so bad
I always wished it at least included the orbiter name in it. But I suppose the orbiter choice could change closer to actual launch.
STS-41B-ATL2
Ouch, shortly after they get standardized and consistent flight numbers, the shuttle program gets cancelled. I guess computer science doesn’t have a monopoly over the difficulty of naming things.
Apollo 14 had a piece of loose solder in the button triggering abort-to-orbit, so it occassionally triggered itself. This wasn't a problem en route to the moon, but the second the descent phase started it would have been a Poisson-timed bomb that would prevent the landing.
There was a bit of memory that could be set to ignore the state of the abort button (this bit was the reason the abort sequence wasn't triggered en route). The problem was this ignore bit was reset by the landing sequence (to allow aborting once landing started), and they did not believe the astronauts would be quick enough to set the bit again before the button shorted out and triggered the abort.
(Ignoring the abort button was fine because an abort could be triggered in the computer instead. Takes a little longer but was determined a better option than scrapping the mission.)
Don Eyles came up with a clever hack. Setting the program state to 71 ("abort in progress") happened to both allow descent to start and prevented the abort button from being effective. So this program state was keyed in just before descent.
The drawback was that it obviously put the computer in an invalid state so some things were not scheduled correctly but Eyles and colleages had figured out which things and the astronauts could start those processes manually.
Then once the computer was in a reasonable state again the ignore abort bit could be set and the program mode set correctly and it was as if nothing had happened.
This was basically a lawn-chair rocket for two which would utilise a disabled LEM's (lunar excursion module) fuel tanks, and would be hand-piloted without any guidance computer to an intercept orbit with the Apollo Command Module, with the hope that a rendezvous and crew transfer could occur within the four-hour window of space-suit oxygen supplies. Given that the CM's orbital period was two hours, this meant at best two chances for a successful intercept.
<https://en.wikipedia.org/wiki/Lunar_escape_systems>
(I'd run across this from the recently submitted MOOSE article, "Man out of space, easiest", a strap-a-foam-mattress-to-your-ass reentry concept: <https://en.wikipedia.org/wiki/MOOSE>.)
It does still leave the system without a means of aborting if the ship's main engines have trouble, although I suppose they do have a good bit of redundancy there.
I'm no expert, so this is a question to confirm my understanding: Starship does have a booster. So, doesn't that make it a dual stage?
https://www.zenger.news/2023/11/27/elon-musk-reveals-simple-...
Even if it's a bit more than doable by hand a ratchet jack should make short work of it.
Plus, you've got to get the whole jacking operation done without damaging any of your main or control thruster rockets, and without tipping past the upright point over to the other side, or just effectively rolling onto an adjacent side.
I wouldn't want to go on a craft where [jacking it back upright] was in the top ten on the list of recovery options to get home.
For a counter, I wouldn't want to be on a lander so fragile[0] that being manipulated upright is infeasible. Something will go wrong, maybe not on that lander, but when it does go wrong it'll have to get duct taped together.
[0]Not just mechanically, but in terms of operation scope. Planning that everything must go perfectly or people die is a recipe for the latter.
(And yes, I've done a fair amount of wrangling vehicles, gear, etc. in snow, dirt, mud, & rocks, and eventually it can often be gotten out. But on a different planet/moon, it really should be not be anything close to a primary option. OTOH, if it's got a set of 6+ pop-out lever-legs to upright itself, tested, etc., that's a different solution)
For comparison, a craft built for earth launch mass fractions probably wouldn't survive falling over in the first place - when that happened to a Falcon 9, the whole rocket simply exploded.
https://xkcd.com/1244/
I don't think that is an assumption you can make. In the worst case scenario the lander lands on the door. In which case the only way to disembark is to lift the lander.
In addition to the obvious, it should be taken into account that the absence of atmosphere makes very difficult to assess distance and scale. Videos of approach seem like a fractal browser.
https://en.wikipedia.org/wiki/STS-1
With both Crew Dragon and Starship, there will have been _many_ successful missions involving un-crewed variants of the spacecraft (Falcon 9 and Cargo Dragon were both well-proven systems before crew was a possibility).
flatten a landing pad
make a moon highway
Moon GPS (2 votes)
stupid-wide landing legs
Dynetics lander
spherical lander
mechanism to push itself into upright
land sideways (ok this was a joke right?)
I couldn't find the relevant XKCD but I remember one along the lines of "why did they just do Y?". I say all of this in jest, as this is what we do here at HN: give our opinion on areas outside our expertise.
Edit: My point being: spacex is already doing it on earth, dealing with stronger gravity and air non linearity.
More seriously, the humans in question were very skilled pilots with huge amounts of general flight experience and specific lunar training; they also had access to hardware that had already been expensively tested in the lunar environment. Neither of those were available to this project.
It probably would have landed upright if the LIDAR worked. It is impressive that it landed as intact as it did
[0] https://arstechnica.com/space/2024/02/it-turns-out-that-odys...
https://en.wikipedia.org/wiki/Mars_Pathfinder#/media/File:Pa...
Having an atmosphere helps, though and that's not available on the Moon.
https://en.wikipedia.org/wiki/Luna_9
It is very odd however to see no mention of the fact that China is about a decade ahead, having completed the Chang'e 3, 4 and 5 missions. which included a successful return of Moon rocks to the Earth. Two more missions are in the works.
The article mentions a Japanese effort, but the omission of China's successes seems fairly deliberate. I wonder what the motivation was?
https://www.space.com/china-new-moon-rover-change-7-mission
Obviously much more complex, but probably cheaper than redundant everything.
https://en.wikipedia.org/wiki/Mars_Pathfinder
It's not only a matter of adding complexity, mass, money, etc. This kind of solution is simple on its face, but it is akin to being ignorant of why a car has windows (for the driver to see), doors (for the passengers to enter/exit) or a cooling system (to not overheat), and then reading that Jeeps (or to be more fair, a new Jeep prototype) are prone to tipping over on a trail after driving 1000km on a highway to a trailhead, and then commenting on how Chrysler should just make a dodecahedron Jeep with 20 wheels instead of 4.
These spacecraft also made it past 99% of the hurdles, the next try will probably make it and overcome the final 1% without starting from scratch and adding all kinds of complexity that are far more likely to add failure.
I work in the industry and opinions are mine and not of my employer.
Obviously N-way redundancy isn't optimal, but it would seem potentially simple if you had a modular design with multiple instances of the same "omniface" component. Perhaps use 1/2 of each redundant face for radiator and 1/2 for solar panel?
I'm not suggesting a rolling design - just one with legs on every side such that if it did tip over (or worst case tumble if landing on an incline) from preferred engine-down orientation then it would not make any difference.
Seeing as Intuitive Machines wanted to avoid deployable (spreadable) legs as a way to achieve a more stable low center of gravity, another alternative would seem to be to shrink the design (lose the height) to better work within the width of the faring, although I don't know how viable that would be given amount of propellant needed for the landing.
"Just put 1/2 radiator and 1/2 solar" and "just put legs on every side" is still really simplistic and nobody even mentioned thermal insulation yet, which is normally a much higher proportion of surface area compared to radiators and solar, at least if all three are on the main spacecraft body. I was being extreme in my last comment to make a point, but you still sound to me like a guy suggesting 20, and now 12 wheels on a Jeep instead of 4.
No - on a lunar lander, not on a jeep. It makes a difference (I'm really not sure why you started talking about Jeeps).
The difference is the cost to test a lunar lander (on the moon) vs cost to test a jeep, and the fact that we're only going to build one lunar lander vs a production run of millions for the jeep.
For the Jeep it's cheap enough to iterate and test (on earth), and end up with an optimized design that has 4 wheels and not 12 or 20. Not only is this relatively cheap to do, but there's a huge incentive to do so since we're going to mass produce the Jeep and would therefore like to reduce build cost and maximize profit.
In contrast, the lunar lander is very expensive to "test" since that means a failed mission to the moon. Maybe that mission costs $5M, so if we fail twice before getting it right we've added $10M to the cost. If we can avoid those failed tests by instead adding $1M of redundant hardware that makes it work first time, then we've saved a lot of money. We're only building one lunar lander, so there's no multiplier in front of that $1M we added to the hardware cost.
Of course it's a more complex design in a way given that you deploy this airbag at low altitude - something else to go wrong. It's interesting though that NASA has used that and the even more complex air-crane technique on Mars both successful first attempt. Even though they seem complex I'd assume NASA determined they were the least-complex approaches that had a high likelihood of success.
Maybe the economics of doing this on the moon, and with a cheaper lander, are different - better to have a simpler system with higher chance of failure, and redo the mission if it fails ?
https://en.wikipedia.org/wiki/G%C3%B6mb%C3%B6c
Doesn't help on the moon, though, where the surface is soft, and may contain boulders.
What if airbags inflated around the lander? The external shape of the lander-with-inflated-airbags and the internal weight distribution could make it self-righting, while shielding the external fixtures and the hull.
Just throwing ideas around and role-playing spacecraft designer.
https://en.wikipedia.org/wiki/Weeble
So then here's my luddite take. Can't they just unfurl wider stabilizers from the legs that increase their footprint? At slow speeds in low gravity it seems like they wouldn't need to be heavy or strong.
So much cheaper just to have an altimeter and no extra stabilizers. Sadly, their altimeter didn’t work. They forgot to turn it on correctly.
Honestly, they’re lucky that it got so close to landing correctly; if it had been going faster the damage would have been even worse than just a crumpled landing leg.
The Apollo LM weighed 15-16.5 metric tons at launch. Not exactly a Mylar balloon.