The article discusses the absolute error coming from RTK systems and claims that it won't be as low as 0.5cm, but surely the relevant metric is relative error, and I can see commercial systems advertising that level of precision.
i.e. the booster doesn't know it's actual position to within 0.5cm but it knows it's position relative to a buoy or the catch arms to that precision.
When Tim of EverydayAstronaut quoted this[1], as I recall the quote was within 0.5cm of the target landing site. So I assumed that to be relative accuracy and not absolute.
Rtk already is 'relative' error- it requires one or more base stations (with either known absolute location or assumed one for relative positioning).
But survey grade gnss is a web of rabbit holes, if you want to get into it.
And there are ways to get sub mm accuracy both relative and absolute, but idk of one that would be quick enough for the required reaction time of dynamic landing via 'catching'.
But multi-centimeter (4-5) that's really easily doable is probably good enough for other systems to take over from.
I'd be very interested in the systems advertising that! I have not seen that even for stationary surveying equipment. I think it's also important to distinguish between RMS error which is often the better topline spec that companies give you, vs the 95% confidence error which is the more relevant one for flight reliability.
In the specific case of a docking-type manoeuvre presumably you only need the highest accuracy when you're getting very close to the target.
No reason you couldn't use RTK GPS for <10cm accuracy for most of the flight, then in the last few meters of landing switch over to to high-precision, short-range tracking - like optically tracking a marker on the grabbing arm.
For other specific cases - like bridge monitoring - there are reports of 2–3 mm precision [1]. Of course, bridge monitoring has quite distinctive requirements; a 5Hz vibration component and a 0.0001 Hz thermal expansion component. So there's a lot of potential to average over lots of readings to reduce noise.
Global lat/long coordinates are defined in terms of coordinate systems like WGS84 or ITRF2020, which are themselves the result of relative measurements between reference stations.
The earth's crust floats on top of liquid rock. This matters at relevant length and time scales; in most places, these effects alone are on the order of millimeters per year. One reason why it's better to use NAD83 over WGS84 in North America is that NAD83 latitudes and longitudes move with the North American plate.
Positions _are_ relative, and the closer you can put your datum, the less drift you'll accumulate.
There is a literal, autistic sense in which you are correct. But there is a practical, pragmatic distinction between measurements that we call absolute versus those we call relative, and pedantic correctness misses the point.
My lawnmower (openmower) can do <2cm accuracy over GNSS. It is absolutely believable they could achieve 0.5cm on GNSS (plus rtk correction data from a fixed base station nearby) alone without measuring any relative distance using other systems.
Well, it isn't "uncorrected". It's just that you don't need any additional hardware other than a second gnss receiver on the base station and some kind of link between them.
GNSS is more than accurate enough once you know all slight errors in satellite orbits and the atmospheric distortions currently affecting the area near the base station and can correct for them.
It's significantly more difficult to actually land a jumbo jet sized rocket booster with that precision than to measure its own relative position. Gerstenmaier was talking about landing accuracy. My guess is that measurement accuracy is a red herring. More likely it was a slip of the tongue (the good man is 70 years old) and he meant to say it landed with a 0.5 meter, not centimeter, accuracy relative to the buoy.
I would have thought the vibration from the engines would produce error of greater that 1/2 cm. Still, it seamed to have worked well. So, there you go.
Armchair aerospaceing here, but it feels like he's a whole class of positioning sensors in this analysis. It seems to be you only need GPS and related absolute positioning systems to get you close to the tower. At that point, what you care about is the relative positioning of the tower and the booster. I would think this can be done very accurately with a host of options: cameras, radar, lasers, ILS style systems, etc, etc.
Not really, the Gyros described in the post are also essential. If you don’t know the attitude of the vehicle, you can’t point the engines in the direction your control algorithm says you should to hit the target.
Edit: I think I misunderstood the comment. Yes, you can use the absolute methods for rough guidance and then use relative positioning for the final approach. The article has a line about why the author doesn’t think that’s likely though.
GNSS RTK is incredibly accurate these days. By the time that you're close enough to the landing zone, you're close enough to get positioning down to centimeters on consumer grade hardware, which the article points out.
The actual question is literal: Can SpaceX land a rocket with sub 1 cm (1/2 cm) accuracy? GNSS RTK can get you down to a couple of centimeters, but getting more granular resolution than this isn't reliably possible with current professional grade technologies.
I'm personally unsure if the military has greater resolution than what's possible with RTK or w.r.t. military use GPS, but I would not be surprised if they did. If that's the case, NASA would most likely have access to it, I would assume. But the article specifically calls this out saying that it's not accurate enough to surpass the resolution of using RTK.
What's really cool about these questions is that the same problem space is applicable to self-driving cars and SLAM, if you're into that sort of thing. Lane detection, etc.
Yes, for the purposes of landing speeds. In fact, at vertical aircraft landing speeds, your time-step to position Δ is more accurate than automotive SLAM.
Depends on what you define as 'traditional': basically if you want positioning information via GNSS, the techniques for getting a better reading a fairly well understood, and it doesn't really matter how good your satellites are, the atmospheric distortion is the issue and you need to model and compensate for it by measuring it with a near-enough base station, using multiple frequencies and constellations, and if you're moving, an IMU to constrain your motion over as long a period as you can to effectively average out the other noise sources. Half a centimeter, given all of the above, is better than what I've seen quoted in the space, but not utterly crazily so.
Right? They land fighter jets on carriers with a light signal that projects out from the ship at a particular angle. It seems very easy to do something like this with some form of electromagnetic radiation. Or have some way for the tower to detect the exact position and communicate with the rocket.
I understand engineering is complicated but this honestly seems like the easiest part of the problem to solve.
It's more likely that SpaceX determined they didn't need super tight tolerances and called it a day.
> SpaceX determined they didn't need super tight tolerances and called it a day
Yup! This is my conclusion in the article - the landing box for the Super Heavy booster is 5x13x18 meters on each side, with 5-15 degrees of angular tolerance in each of the vehicle axes. So the margins are big enough that you don't need millimeter level precision for the rocket position.
My take is that it would probably be possible with enough effort, but there isn't an easy solution. And if you don't need it then the best part is no part. :)
Carrier landings are accomplished using a combination of indications, and the meatball is only one of three primary tools. If you are not flying an on-speed angle of attack AND lined up on the centerline of the landing area, the meatball position is invalid to a degree proportional to the degree to which those other inputs are off.
The meatball Fresnel lens is canted slightly side-to-side, and only places the hook in the right spot at a given angle of attack. Which is a design compromise necessitated by having to allow multiple types of aircraft with multiple hook-to-eye distances to land on the same aircraft carrier while using a visual input in one location (the cockpit) to properly place a device in another location (the hook point) with high precision.
Source: I've done it.
So just as it is not "very easy" to trap on board the boat with "just" a light signal, I would assume landing a building-sized booster has a similar if not bigger list of potential "gotchas."
If they did that and hit them consistently.. talk about rubbing salt in the wounds of the rest of the industry. It would be like a sports team running up the scoreboard on an obviously beaten opponent. Super heavy hitting X's painted on the chopsticks right in the middle would be border line unsportsman-like heh.
You can do even better with radar. If you place a set of radar reflectors around the tower at known locations, then you can detect them from the booster and triangulate the distances to a precise position. Plus, radar gives you relative velocities, so your speed and roll rate estimatimates get even more precise. I bet you could get down to millimeters with a setup like this.
Falcon-9 uses radar altimeters for determining vertical "distance to go" during landing.
While a sideways position error of even ten meters is not fatal, it is critical for the rocket to be quite close to zero altitude when deceleration brings the velocity to zero. (Any residual error must be dealt with by the shock absorbers, and their capability is modest.)
The tricky parts (that we don't really know as non-SpaceX employees) are:
- how accurate is the clock onboard the satellites? Given that it's likely an OFDM signal the timing is probably pretty good, but given that they're launching zillions of them they probably don't all have atomic clocks onboard
- how accurately is SpaceX tracking their orbits? Kind of a similar answer here... they're doing beamforming to the ground terminals, so it has to be pretty good but we don't really know how good.
- how many SVs are actually visible at a time? We need a minimum of four but the more the better. If there's lots visible we can somewhat work around the first two issues statistically but if there's a limited number than the orbit and clocks need to be super accurate.
Also—how accurately do those satellites track their own position? Unlike the high-orbit GNSS constellations, LEO satellites would bounce around a bit from orbit to orbit, as they're relatively close to the earth and sensitive to uneven distributions of mass. I don't know the exact magnitudes, but I understand they're large by GNSS standards.
Starlink satellites use on-board GPS receivers for extremely accurate (centimeter level) measurements of their position. The orbits which SpaceX reports to the world (for collision avoidance) are based on these measurements.
I think they were simply speaking of the final pads the rocket rests on top of the chopsticks have 5cm of error either direction.
But judging from the bouncing the rocket did when in the chopsticks the error for positioning into the initial catch position is much larger in all directions. The chopsticks coming closed around the rocket do the heavy lifting for final alignment to that 5cm I imagine.
Landing on the grid fins would be a really bad idea. Even though they're car-sized, they're not load bearing and "only" made of steel (not titanium etc. .. just yet). Starship's Raptors blast during hot staging is enough to bend them on the top. https://www.reddit.com/r/SpaceXLounge/comments/1g3bi7s/grid_...
I actually think there is some old Starbase tour interview where a SpaceX guy implied it was Musk's idea, though I could be misremembering. Catching the booster kind of makes sense, since they needed the tower arms anyway for stacking and unstacking.
Interesting, are there more than 4? because I was also amazed that the rocket was rotated at exactly the right angle to be caught by them. But maybe that is the 'easier' challenge when you are hovering with such accuracy.
I keep finding myself watching the catch every few days, and it does not tire to impress.
> Bill likely misspoke or was talking about control error.
Mixing up control errors with absolute errors is a very common form of miscommunication in robotics.
I work with relatively big robots and often my colleagues would say something like this "During the test we had 0.5m cross track error, so we did X, Y, Z ...".
And I always ask them for clarification. Were they looking at the robot and seeing that it is half a meter off where it should be, or were they looking at a screen and seeing that the robot thinks it is half a meter off from where it wants to be? Because those are two very different situations. And both can be described with the same words. (And sometimes it can be both, or just one of them.)
The robot knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation.
This voice sounds like something that Mark Farina should be dubbing into his next album. But it's the first time I've heard this bit. Where did it come from? Is this a classic in engineering circles of some shit Rockwell actually sold to the military?
It's from an old air force training video. Best guess I'be heard it that it's an unsuccessful attempt to explain Kalman filters (or something similar) in layman's terms.
It's definitely floated around for a while, but it grew in popularity in the past few years.
Control error is defined as the difference between desired value and measured value. So this is pretty good?
Even if they use some crude method to obtain position (e.g. gps), they can still easily refine that using e.g. triangulation using cameras around the landing platform.
Not sure what you are talking about. If you are asking if 0.5cm is good controller error for an orbital class launcher on landing? Yes, it is extremely good. Without doubt.
If you are asking about my tangential story where there is confusion between total error vs controller error then no, it is not good. Confusion is never good. Especially if the system is not within the total error budget. Because to improve it you need to know if you are dealing with measurement error or controller error.
> Even if they use some crude method to obtain position (e.g. gps), they can still easily refine that using e.g. triangulation using cameras around the landing platform.
Sure. I doubt that their total error is within 0.5cm, but both of their landings were extremely succesfull.
From what I understand this is the primary difference and problem with Falcon 9 vs. Super Heavy.
From the Article:
"Why can't SpaceX do a catch with a Falcon 9?
-It does not have separate landing propellant tanks, so propellant slosh will disturb its trajectory. The Super Heavy booster has dedicated central header tanks for landing propellant, so there should be minimal propellant slosh to disturb the vehicle attitude.
-It lands with a single engine which cannot throttle low enough to hover the vehicle, and as such must perform a “hoverslam” maneuver to bring the vehicle to a stop right on the ground. While the Super Heavy booster must perform most of a hoverslam maneuver to slow down just before coming in to the tower, it can hover for the final fine positioning.
-Because it lands with a single engine, roll control is minimal close to touchdown when the airspeed is low and the grid fins can impart minimal torque, and is limited to its weaker cold-gas thrusters. The Super Heavy booster can control roll with its 3 engines all the way to the ground.
-Falcon 9 has no engine-out capability for landing. SpaceX has not confirmed it for the Super Heavy booster, but I believe one engine out is likely possible (more on this later).
-It is smaller with a lower moment of inertia. Rockets get more stable and easier to control the larger they are, much like it’s easier to balance a broom on your finger than a pencil.
-It is smaller, and so thanks to the cubed-square law has a higher area:mass ratio. This means that it will be more affected by wind gusts that might blow it off course."
You can see the landing arms adjusting in the video as well.
His choice of stainless steel is panning out well here - I doubt if aluminum or composite body structure would hold up as well to the "grab" forces from even minor misalignment. A composite structure would likely be entirely compromised by a big scrape.
It would be interesting to see a test where the landing speeds were deliberately too high - how much deceleration can the arms handle safely?
I think the chopstick mechanism is probably the best possible catch mechanism for such a tall object. The booster will be suspended from the top, which means the booster isn't subject to tipover as it would be if landing legs were involved. We've already seen this many times in the Falcon 9 booster series.
I can't see chopsticks ever working from a droneship, though - too much induced rotation for chopsticks to compensate.
As an alternative to chopsticks, a catch 'sleeve' might be possible, though it would magnify alignment errors considerably.
0.5cm isn't even well defined, considering that the rocket itself is probably out of round by larger tolerances, not to mention issues of thermal expansion.
is "surface amplitude" vibration? I could see vibration being graphed over time as a wave with an amplitude and frequency. I can't really comprehend the size of super heavy combined with the energy density of just one turbo pump on one raptor engine (let alone 33x2) and then the precision of control needed to catch the whole thing with chopsticks. Not many things do i admit are just beyond me period but this is for sure.
/raptor 3 pumps look like you could hold them in your hands but iirc they deliver over 100k horsepower each.
It's probably defined by the catch points as that's what matters: whether the catch points end up where they belong (good) or not (disaster). The catch points are not "out of round".
I should have mentioned that the catch points are on one long bar that goes atop the booster. They are not simply attached to the sides of the booster. Booster shell flexing does not affect the catch points at all.
I would suspect it's some average of many points that the control systems try to track, because the rocket is not a point, and the press wants to quote a single easy number.
I'd expect that the rocket has a ton of sensors, and a ton of passive and semi-active tracking devices all over the body.
E.g. I'd put a bunch of NFC-type responders in a number of key positions, responding at different frequencies. Then a typical sweeping-frequency radar pulse would activate them all, and the response time and the Doppler shift would tell about positions and speeds of many points on the rocket. I'd do a similar thing with reflectors and IR/optical tracking.
All these points should follow some reasonable trajectory for some point the top of the rocket, near the chopsticks, would move towards some desired catch location point. Probably this motion is where "with precision of whatever cm" relates to.
Well, to be fair, the engineers working on this system have been thinking about the accuracy this whole time even if you haven't. Even on the first go, it still had to land on a relatively small pad floating in the ocean.
As impressive as the space efforts in the 60s and 70s were, I've often thought that they were a false start created by a war-like impetus to show off. Tech-wise, we really weren't ready for a space age. The sort of control systems that make this sort of outcome possible haven't been around for all that long, really, especially if you mark them from being economical and not just "it technically existed in a lab somewhere". Plus if you really dig into how these rockets are built and maintained, you see a lot of other technologies that have not been around for that many decades, like, practical and reliable 3D printing, and computing simulations that have more computational power per second than the entire computing world could scrape together in a year in the 1960s, and those are just the highlights, not the exhaustive list.
A lot of people are like "we got to the moon in the 1960s, where's the progress we should have had since then?" but I see the 1960s as the bizarre exception rather than the thing that should be used to set the rule. There was no way the space age was going to happen then, in an era where you're almost sitting there counting each bit of RAM you can afford to send into space. The true Space Age is just dawning now, and it's still early in the dawn; we still have to have massive international cooperation to put a single space station up, we can't do something as basic as refuel in orbit, we just barely started having people in space for commercial rather than governmental reasons... it's just the beginning.
I think the 60’s showed how much humans can achieve in terms of innovating with very little (in terms of tech). Now, we’re seeing how much can be achieved with a whole lot more. And, I agree, the space age really does feel like it’s only just heating up. Very exciting time!
I think it's not so much that we weren't ready for a space age tech-wise, but that the the reason we have so much of our technology today is because of investments made in the 1960s. NASA had basically unlimited money to throw at every technical challenge in the way of landing a human on the moon.
The apollo program drove the need for more computational power, more memory, better guidance and navigation and control systems, better materials, experiments to better understand many phenomena, etc. And after the apollo program ended, the contractors that developed those technologies on NASA contracts could just commercialize them. And the data from experiments, on materials, aerodynamics, combustion, and so on, that is publicly available has made engineering so much cheaper and easier.
> There was no way the space age was going to happen then, in an era where you're almost sitting there counting each bit of RAM you can afford to send into space.
And yet, they got to space. Better computers are not the solution to every problem. And it wasn't a false start. We are already in a space age, we have been for quite a while. It didn't stop at Apollo. We have satellites for the military, TV, weather, various forms of communication, navigation (GPS...), telescopes, space stations, probes and rovers. We do science, commercial and government operations. Starlink is great, but it is just the continuity of all the space communication abilities we developed over the years.
I think computers are not what will enable the "true space age". Sure, they help, and SpaceX, if successful with their Starship will certainly be a big advance, but I think that we are missing a key ingredient to reach the "true space age" and it is nuclear power. Starship maybe could get us a settlement on Mars with hundreds if not thousands of launches and refueling missions. Project Orion was to launch an entire colony in one go, return trip included. Even Saturn was considered feasible. Project Orion is mad, but it goes to show how limiting chemical rockets are compared to nuclear.
And it is something we probably could have done already, without modern computers and 3D printing, if we wanted to. It is maybe a good thing that we didn't though. Spreading radioactive material in the atmosphere and mass producing thermonuclear bombs is kind of scary.
Yes, that seems like a very big hand-wavy assumption. This paragraph is quite...something:
> Could you use other real-time distance measurements like laser rangefinding or visual processing? I don’t think so – the surface of the vehicle is too irregular to get a reliable fix point, especially while it is moving, and these are vulnerable to smoke/fog/gas/ambient lighting. Technologies like Ultra Wideband are vulnerable to multipath reflections and attenuation by the booster’s steel walls, and aren’t more accurate than RTK anyway.
That is not exactly an exhaustive list of methods to locate an object.
I have no idea whether 0.5 cm precision is feasible or even needed, but this part felt a bit off.
As the article explains, with a well designed procedure, the required navigation accuracy is quite modest. Even the latest consumer IMUs and GPS would do, and SpaceX is using even slightly more accurate "tactical grade" units, typical for all launch vehicles.
Good article. It is nice how it goes through all the points systematically.
It talks about how real time control system algorithms work with algorithms like like PID and MPC. I assume the SpaceX solution is likely one of the most advanced control systems in the world.
It is great that the author of the video helps to introduce the topic to the general audience. This is a vast and a fascinating subject with lots of material available for a further more systematic study.
It becomes more technical, if one is specifically interested in the methods used by SpaceX. But there are some overviews that describe the general idea. Here is one presentation by Behçet Açıkmese: https://nescacademy.nasa.gov/video/eda2b96bddf945629be2c9d2e...
Note that before joining SpaceX to lead their autonomous landing software development, Lars Blackmore worked at JPL, where together with Behçet Açıkmese they developed the autonomous precision landing algorithms based on real time optimization methods. So, even though SpaceX has undoubtedly developed additional nuances to match their needs and capabilities, they were building on this prior work at JPL.
It's interesting that NASA, despite working on control algorithms, apparently didn't consider using them for building a reusable rocket. (SLS isn't planned to do a propulsive landing, and neither was its predecessor Ares V.) Though they probably did use them to some extent for the "sky crane": https://en.wikipedia.org/wiki/Sky_crane_(landing_system)
Making any kind of a rocket that works is already nontrivial (once one goes beyond Estes models), but in terms of complexity of the challenge, making some kind of a hopper that goes up and down is a task that ambitious amateur groups with a few members and under $1M in funding were demonstrating even before SpaceX was founded. There is no need for a highly efficient engine, there is no need for lightweight structures. Many other concerns, such as low frequency structure oscillations, aerodynamics, etc are practically nonexistent.
Making a space launch vehicle is a task for a group with at least x100 more resources and experience. Reaching orbital velocity is pretty hard. Most startups do not succeed.
Making a space launch vehicle which does not spend the fuel entirely, while carrying extra hardware to also land after the launch is another step up in how hard this is. Very serious institutions worked on this problem since 1970s but lots and lots of people were skeptical. Shuttle was impressive, but also very expensive. Then SpaceX has shown that it was not only possible, but even practical to make ordinary rockets reusable. That was amazing. Even now, almost a decade later, nobody has shown anything like that -- though a number of Chinese startups are working on it.
The accuracy that counts is for the gap between the arms and the booster, which could well be 1cm since they can measure that with sensors as the booster descends between the arms, and the arms can be controlled accurately.
> Half a centimeter landing accuracy is not possible, and Bill likely misspoke or was talking about control error.
Maybe the 1/2 cm accuracy refers to the final position of the booster's catch points on the arms after they've closed, after the booster's engines are off, and after the booster settled, and maybe they mean lateral accuracy. I would forgive them for that because that's the accuracy that actually matters here.
If the catch points were off then that might spell disaster, so the catch points' landing accuracy including the help of the catch arms is what matters.
The catch/lifting points may look small, but actually protrude from the side of the booster by 2-3 feet. Note that the booster is actually in a hover at the point the arms close in to touch it, so as long as it's vertical rotational axis is right (there are only 2 pins - one on either side), the positioning of the pins on the catching arms is basically guaranteed.
Right, so if SpaceX meant that they had an error of only .5cm maybe they meant that the error on the booster rotation angle was small enough to produce only a .5cm error at the catch points. Since they weren't specific, it's hard to know what they meant.
The booster rotation angle error and the catch point placement error were much too small to detect with the naked eye on the published videos. Every other measure of accuracy was clearly within tolerances -- and also hard to discern with the naked eye.
As amazing as .5cm accuracy sounds, if SpaceX meant catch point placement error, then it's quite as impressive because that only implies everything was within tolerance _and_ only the booster rotation angle error need have been impressively near-zero measure. That's... still amazing, honestly. If you can get the booster rotation angle error near zero then you can get the other errors way down too.
Most of this article feels like it's discussing irrelevant methods, you only need GPS to get it close (well for what they're doing they don't need GPS at all, though I'm sure it's used), we have much much more accurate ways of measuring the positions of things from a fixed reference point, 0.5cm deviation on your positional measurement is trivially achievable with optical systems. Why is the author spending paragraphs discussing IMU accuracy when we're trying to line up a rocket with a tower. You care about the rocket's relative position to the tower, you can put your measurement equipment on the tower, you don't need to worry about how accurate your accelerometers are.
I assume they are doing something much more clever/hardened, but you can trivially achieve much greater spatial accuracy with a Vive Tracking Puck for like $100.
Bill Gerstenmaier was talking about the flight test 4 landing accuracy, which landed on the open sea in the Gulf of Mexico, not on the tower like the recent test flight 5. The only thing nearby was a buoy. I'm pretty certain it didn't have advanced laser systems.
There’s a big cross in the middle of the landing pad that you’re trying to aim for - you don’t need advanced laser systems to get an accurate fix on where the landing pad is from the rocket - or where the rocket is form the landing pad for that matter.
Certainly lab equipment can measure distances well below 1 um fairly easily, I could manage 1 um in my garage. The issue is that the conditions at time of catch are VERY dynamic and not at all lab-like.
Your positioning system needs to acquire a fix at least 100m out in variable atmospheric conditions on a rocket undergoing heavy acceleration and dumping all kinds of heat, smoke and vibrations into itself, the atmosphere, and everything around it.
In addition having a fix on your tracking device is only half the game, not you have to figure out where the rest of the rocket is in relation to your tracking device. Which again, vibrations, temperature and manufacturing all have an effect.
So while yet, a vive tracking puck isn't entirely unlike the workable solution it is also entirely unsuitable as a solution and should not be used as a baseline to measure off of.
So? Yeah it's a challenging environment, we know that. My point is that the default way to solve this problem is to track your object from your reference point.
If it's positioning relative to the chopsticks, I'm sure it's possible to know where you are within a centimeter, even with all the rocket exhaust flying around. That's what DGPS is all about.
It's still wildly un-nerving to me that there's no publicly stated option other than the chopsticks for landing(edit: some future passenger craft). Imagine if you've got enough fuel to avoid slamming into the ground, and a nice big ocean, or a lake sufficiently deep... couldn't a water landing happen and let future passengers survive?
This is not the passenger vehicle. That's the second stage, this is the first stage booster. They've also landed the second stage but that did a water landing from orbit which in many ways is even more impressive.
I found this video from the perspective of a landing pin interesting, especially when played back at low speed: https://youtu.be/ExV6PHRM8eI?t=17
You can see the arm comes in, then there's some side-to-side bounce (not sure how much is the rocket bouncing off vs. the arm fine-tuning its position). Just after contact seems to be made, and before the shock absorbing (or yaw-correcting) pistons drop much, there's a large flash from the engine. Is that a characteristic of engine shutoff, or was there a last-second "hover" push just before shutoff and drop? I wonder how much force the arms felt.
Another perspective showing both arms, and (as mentioned in the article) how the left one adjusted more significantly at first: https://youtu.be/JlcrNakUGVs?t=3
172 comments
[ 3.1 ms ] story [ 96.3 ms ] threadi.e. the booster doesn't know it's actual position to within 0.5cm but it knows it's position relative to a buoy or the catch arms to that precision.
[1]: https://www.youtube.com/watch?v=pAPt5vbr-YU (don't recall timestamp, sorry)
But survey grade gnss is a web of rabbit holes, if you want to get into it.
And there are ways to get sub mm accuracy both relative and absolute, but idk of one that would be quick enough for the required reaction time of dynamic landing via 'catching'.
But multi-centimeter (4-5) that's really easily doable is probably good enough for other systems to take over from.
Or in fact, you need even better than that, since you don't want your whole error budget used up by the GNSS system.
No reason you couldn't use RTK GPS for <10cm accuracy for most of the flight, then in the last few meters of landing switch over to to high-precision, short-range tracking - like optically tracking a marker on the grabbing arm.
For other specific cases - like bridge monitoring - there are reports of 2–3 mm precision [1]. Of course, bridge monitoring has quite distinctive requirements; a 5Hz vibration component and a 0.0001 Hz thermal expansion component. So there's a lot of potential to average over lots of readings to reduce noise.
[1] https://www.sciencedirect.com/science/article/abs/pii/S02632...
The earth's crust floats on top of liquid rock. This matters at relevant length and time scales; in most places, these effects alone are on the order of millimeters per year. One reason why it's better to use NAD83 over WGS84 in North America is that NAD83 latitudes and longitudes move with the North American plate.
Positions _are_ relative, and the closer you can put your datum, the less drift you'll accumulate.
GNSS is more than accurate enough once you know all slight errors in satellite orbits and the atmospheric distortions currently affecting the area near the base station and can correct for them.
Edit: I think I misunderstood the comment. Yes, you can use the absolute methods for rough guidance and then use relative positioning for the final approach. The article has a line about why the author doesn’t think that’s likely though.
The actual question is literal: Can SpaceX land a rocket with sub 1 cm (1/2 cm) accuracy? GNSS RTK can get you down to a couple of centimeters, but getting more granular resolution than this isn't reliably possible with current professional grade technologies.
I'm personally unsure if the military has greater resolution than what's possible with RTK or w.r.t. military use GPS, but I would not be surprised if they did. If that's the case, NASA would most likely have access to it, I would assume. But the article specifically calls this out saying that it's not accurate enough to surpass the resolution of using RTK.
What's really cool about these questions is that the same problem space is applicable to self-driving cars and SLAM, if you're into that sort of thing. Lane detection, etc.
But in realtime? (single-digit second latency, at least)
I understand engineering is complicated but this honestly seems like the easiest part of the problem to solve.
It's more likely that SpaceX determined they didn't need super tight tolerances and called it a day.
Yup! This is my conclusion in the article - the landing box for the Super Heavy booster is 5x13x18 meters on each side, with 5-15 degrees of angular tolerance in each of the vehicle axes. So the margins are big enough that you don't need millimeter level precision for the rocket position.
This is the part I question though. Seems like an org as well motivated as SpaceX could easily solve that if it was necessary.
The meatball Fresnel lens is canted slightly side-to-side, and only places the hook in the right spot at a given angle of attack. Which is a design compromise necessitated by having to allow multiple types of aircraft with multiple hook-to-eye distances to land on the same aircraft carrier while using a visual input in one location (the cockpit) to properly place a device in another location (the hook point) with high precision.
Source: I've done it.
So just as it is not "very easy" to trap on board the boat with "just" a light signal, I would assume landing a building-sized booster has a similar if not bigger list of potential "gotchas."
There have been many airplane crashes because of sudden unexpected wind changes while landing.
While a sideways position error of even ten meters is not fatal, it is critical for the rocket to be quite close to zero altitude when deceleration brings the velocity to zero. (Any residual error must be dealt with by the shock absorbers, and their capability is modest.)
The tricky parts (that we don't really know as non-SpaceX employees) are:
- how accurate is the clock onboard the satellites? Given that it's likely an OFDM signal the timing is probably pretty good, but given that they're launching zillions of them they probably don't all have atomic clocks onboard
- how accurately is SpaceX tracking their orbits? Kind of a similar answer here... they're doing beamforming to the ground terminals, so it has to be pretty good but we don't really know how good.
- how many SVs are actually visible at a time? We need a minimum of four but the more the better. If there's lots visible we can somewhat work around the first two issues statistically but if there's a limited number than the orbit and clocks need to be super accurate.
Impossible apparently.
But judging from the bouncing the rocket did when in the chopsticks the error for positioning into the initial catch position is much larger in all directions. The chopsticks coming closed around the rocket do the heavy lifting for final alignment to that 5cm I imagine.
https://www.youtube.com/watch?v=ExV6PHRM8eI
I actually think there is some old Starbase tour interview where a SpaceX guy implied it was Musk's idea, though I could be misremembering. Catching the booster kind of makes sense, since they needed the tower arms anyway for stacking and unstacking.
I keep finding myself watching the catch every few days, and it does not tire to impress.
Mixing up control errors with absolute errors is a very common form of miscommunication in robotics.
I work with relatively big robots and often my colleagues would say something like this "During the test we had 0.5m cross track error, so we did X, Y, Z ...".
And I always ask them for clarification. Were they looking at the robot and seeing that it is half a meter off where it should be, or were they looking at a screen and seeing that the robot thinks it is half a meter off from where it wants to be? Because those are two very different situations. And both can be described with the same words. (And sometimes it can be both, or just one of them.)
It's definitely floated around for a while, but it grew in popularity in the past few years.
This always stuck out in an otherwise excellent bit, because you should definitely _not_ be taking the absolute value of your control error.
Control error is defined as the difference between desired value and measured value. So this is pretty good?
Even if they use some crude method to obtain position (e.g. gps), they can still easily refine that using e.g. triangulation using cameras around the landing platform.
Not sure what you are talking about. If you are asking if 0.5cm is good controller error for an orbital class launcher on landing? Yes, it is extremely good. Without doubt.
If you are asking about my tangential story where there is confusion between total error vs controller error then no, it is not good. Confusion is never good. Especially if the system is not within the total error budget. Because to improve it you need to know if you are dealing with measurement error or controller error.
> Even if they use some crude method to obtain position (e.g. gps), they can still easily refine that using e.g. triangulation using cameras around the landing platform.
Sure. I doubt that their total error is within 0.5cm, but both of their landings were extremely succesfull.
From the Article:
"Why can't SpaceX do a catch with a Falcon 9?
-It does not have separate landing propellant tanks, so propellant slosh will disturb its trajectory. The Super Heavy booster has dedicated central header tanks for landing propellant, so there should be minimal propellant slosh to disturb the vehicle attitude.
-It lands with a single engine which cannot throttle low enough to hover the vehicle, and as such must perform a “hoverslam” maneuver to bring the vehicle to a stop right on the ground. While the Super Heavy booster must perform most of a hoverslam maneuver to slow down just before coming in to the tower, it can hover for the final fine positioning.
-Because it lands with a single engine, roll control is minimal close to touchdown when the airspeed is low and the grid fins can impart minimal torque, and is limited to its weaker cold-gas thrusters. The Super Heavy booster can control roll with its 3 engines all the way to the ground.
-Falcon 9 has no engine-out capability for landing. SpaceX has not confirmed it for the Super Heavy booster, but I believe one engine out is likely possible (more on this later).
-It is smaller with a lower moment of inertia. Rockets get more stable and easier to control the larger they are, much like it’s easier to balance a broom on your finger than a pencil.
-It is smaller, and so thanks to the cubed-square law has a higher area:mass ratio. This means that it will be more affected by wind gusts that might blow it off course."
That is what they told us in missile maintenance school.
And gyros have gotten a lot better. Especially if you're throwing money at the issue like you know those folks are.
His choice of stainless steel is panning out well here - I doubt if aluminum or composite body structure would hold up as well to the "grab" forces from even minor misalignment. A composite structure would likely be entirely compromised by a big scrape.
It would be interesting to see a test where the landing speeds were deliberately too high - how much deceleration can the arms handle safely?
I think the chopstick mechanism is probably the best possible catch mechanism for such a tall object. The booster will be suspended from the top, which means the booster isn't subject to tipover as it would be if landing legs were involved. We've already seen this many times in the Falcon 9 booster series.
I can't see chopsticks ever working from a droneship, though - too much induced rotation for chopsticks to compensate.
As an alternative to chopsticks, a catch 'sleeve' might be possible, though it would magnify alignment errors considerably.
/raptor 3 pumps look like you could hold them in your hands but iirc they deliver over 100k horsepower each.
I'd expect that the rocket has a ton of sensors, and a ton of passive and semi-active tracking devices all over the body.
E.g. I'd put a bunch of NFC-type responders in a number of key positions, responding at different frequencies. Then a typical sweeping-frequency radar pulse would activate them all, and the response time and the Doppler shift would tell about positions and speeds of many points on the rocket. I'd do a similar thing with reflectors and IR/optical tracking.
All these points should follow some reasonable trajectory for some point the top of the rocket, near the chopsticks, would move towards some desired catch location point. Probably this motion is where "with precision of whatever cm" relates to.
Anybody else thinking this quite the time to be alive?
A lot of people are like "we got to the moon in the 1960s, where's the progress we should have had since then?" but I see the 1960s as the bizarre exception rather than the thing that should be used to set the rule. There was no way the space age was going to happen then, in an era where you're almost sitting there counting each bit of RAM you can afford to send into space. The true Space Age is just dawning now, and it's still early in the dawn; we still have to have massive international cooperation to put a single space station up, we can't do something as basic as refuel in orbit, we just barely started having people in space for commercial rather than governmental reasons... it's just the beginning.
The apollo program drove the need for more computational power, more memory, better guidance and navigation and control systems, better materials, experiments to better understand many phenomena, etc. And after the apollo program ended, the contractors that developed those technologies on NASA contracts could just commercialize them. And the data from experiments, on materials, aerodynamics, combustion, and so on, that is publicly available has made engineering so much cheaper and easier.
And yet, they got to space. Better computers are not the solution to every problem. And it wasn't a false start. We are already in a space age, we have been for quite a while. It didn't stop at Apollo. We have satellites for the military, TV, weather, various forms of communication, navigation (GPS...), telescopes, space stations, probes and rovers. We do science, commercial and government operations. Starlink is great, but it is just the continuity of all the space communication abilities we developed over the years.
I think computers are not what will enable the "true space age". Sure, they help, and SpaceX, if successful with their Starship will certainly be a big advance, but I think that we are missing a key ingredient to reach the "true space age" and it is nuclear power. Starship maybe could get us a settlement on Mars with hundreds if not thousands of launches and refueling missions. Project Orion was to launch an entire colony in one go, return trip included. Even Saturn was considered feasible. Project Orion is mad, but it goes to show how limiting chemical rockets are compared to nuclear.
And it is something we probably could have done already, without modern computers and 3D printing, if we wanted to. It is maybe a good thing that we didn't though. Spreading radioactive material in the atmosphere and mass producing thermonuclear bombs is kind of scary.
There are so many other methods that the lander can use to know where the tower is.
> Could you use other real-time distance measurements like laser rangefinding or visual processing? I don’t think so – the surface of the vehicle is too irregular to get a reliable fix point, especially while it is moving, and these are vulnerable to smoke/fog/gas/ambient lighting. Technologies like Ultra Wideband are vulnerable to multipath reflections and attenuation by the booster’s steel walls, and aren’t more accurate than RTK anyway.
That is not exactly an exhaustive list of methods to locate an object.
I have no idea whether 0.5 cm precision is feasible or even needed, but this part felt a bit off.
Good article. It is nice how it goes through all the points systematically.
https://youtube.com/watch?v=QHikx6kVvAo
It talks about how real time control system algorithms work with algorithms like like PID and MPC. I assume the SpaceX solution is likely one of the most advanced control systems in the world.
It becomes more technical, if one is specifically interested in the methods used by SpaceX. But there are some overviews that describe the general idea. Here is one presentation by Behçet Açıkmese: https://nescacademy.nasa.gov/video/eda2b96bddf945629be2c9d2e...
Note that before joining SpaceX to lead their autonomous landing software development, Lars Blackmore worked at JPL, where together with Behçet Açıkmese they developed the autonomous precision landing algorithms based on real time optimization methods. So, even though SpaceX has undoubtedly developed additional nuances to match their needs and capabilities, they were building on this prior work at JPL.
Making a space launch vehicle is a task for a group with at least x100 more resources and experience. Reaching orbital velocity is pretty hard. Most startups do not succeed.
Making a space launch vehicle which does not spend the fuel entirely, while carrying extra hardware to also land after the launch is another step up in how hard this is. Very serious institutions worked on this problem since 1970s but lots and lots of people were skeptical. Shuttle was impressive, but also very expensive. Then SpaceX has shown that it was not only possible, but even practical to make ordinary rockets reusable. That was amazing. Even now, almost a decade later, nobody has shown anything like that -- though a number of Chinese startups are working on it.
This talk was given at NVIDIA GTC 2015 and inspired me to go into manufacturing
https://www.youtube.com/watch?v=vYA0f6R5KAI
title: "GPUs To Mars: Full Scale Simulation of SpaceX's Mars Rocket Engine"
Maybe the 1/2 cm accuracy refers to the final position of the booster's catch points on the arms after they've closed, after the booster's engines are off, and after the booster settled, and maybe they mean lateral accuracy. I would forgive them for that because that's the accuracy that actually matters here.
If the catch points were off then that might spell disaster, so the catch points' landing accuracy including the help of the catch arms is what matters.
The booster rotation angle error and the catch point placement error were much too small to detect with the naked eye on the published videos. Every other measure of accuracy was clearly within tolerances -- and also hard to discern with the naked eye.
As amazing as .5cm accuracy sounds, if SpaceX meant catch point placement error, then it's quite as impressive because that only implies everything was within tolerance _and_ only the booster rotation angle error need have been impressively near-zero measure. That's... still amazing, honestly. If you can get the booster rotation angle error near zero then you can get the other errors way down too.
I assume they are doing something much more clever/hardened, but you can trivially achieve much greater spatial accuracy with a Vive Tracking Puck for like $100.
Looking at the image I can see a dark device on top of a mast that could be anything electronic.
Your positioning system needs to acquire a fix at least 100m out in variable atmospheric conditions on a rocket undergoing heavy acceleration and dumping all kinds of heat, smoke and vibrations into itself, the atmosphere, and everything around it.
In addition having a fix on your tracking device is only half the game, not you have to figure out where the rest of the rocket is in relation to your tracking device. Which again, vibrations, temperature and manufacturing all have an effect.
So while yet, a vive tracking puck isn't entirely unlike the workable solution it is also entirely unsuitable as a solution and should not be used as a baseline to measure off of.
It's still wildly un-nerving to me that there's no publicly stated option other than the chopsticks for landing(edit: some future passenger craft). Imagine if you've got enough fuel to avoid slamming into the ground, and a nice big ocean, or a lake sufficiently deep... couldn't a water landing happen and let future passengers survive?
You can see the arm comes in, then there's some side-to-side bounce (not sure how much is the rocket bouncing off vs. the arm fine-tuning its position). Just after contact seems to be made, and before the shock absorbing (or yaw-correcting) pistons drop much, there's a large flash from the engine. Is that a characteristic of engine shutoff, or was there a last-second "hover" push just before shutoff and drop? I wonder how much force the arms felt.
Another perspective showing both arms, and (as mentioned in the article) how the left one adjusted more significantly at first: https://youtu.be/JlcrNakUGVs?t=3
https://nextbigfuture.s3.amazonaws.com/uploads/2023/12/Scree...