SpaceX Is Insane! Starship Flight 4 Campaign Continues! Mechazilla Catch Problems!
Vložit
- čas přidán 2. 05. 2024
- Secure your privacy with Surfshark! Enter coupon code FELIX for an extra 4 months free at surfshark.deals/felix
Why did SpaceX ditch the legs? The advantages of catching Starship!
Ship 30 rolls out for a static fire! Booster 13 undergoes cryo testing! Can SpaceX hit 144 launches in 2024? And Rocket Factory Augsburg is getting close to launching RFA One!
SkyFi: app.skyfi.com/explore
#SpaceX #starship #elonmusk #starbase
▶️Patreon: www.wai.to/patreon
▶️WAI Hub: www.whataboutit.space
▶️Redline Helicopters: wai.to/redlineheli
▶️Merch Store US: whataboutit.myspreadshop.com/
▶️Merch Store EU: whataboutit.myspreadshop.de/
▶️WAI Orbit: In collaboration with our friends at SkyFi: / skyfiapp
▶️X (Twitter): www.x.com/FelixSchlang
▶️Instagram: / felixschlang
▶️Facebook: / waispace
▶️WAI Spotify Playlist: spoti.fi/39tmULH
Editing: John Young, Alex Potvin, Stefanie Schlang
Photography: John Cargile, John Winkopp & Stefanie Schlang
3D Animation: Voop3D
Script & Research: Eryk Gawron, Oskar Wrobel, Felix Schlang
LIVE Production: Astro Roadie
Host: Felix Schlang
Production: Stefanie & Felix Schlang
Graphics & Media Processing: Jonathan Heuer, Felix Schlang
Credit:
⭐SpaceX
⭐NASA
⭐Rykllan on X: / _rykllan
⭐SpaceX 3D Creation Eccentric on X: / bl3d_eccentric
⭐SpaceX 3D Creation Eccentric on CZcams: / spacex3dcreationeccentric
⭐TijnM on X: / m_tijn
⭐TijnM on CZcams: / @tijn_m
⭐Jessica Kirsh on CZcams: / @jessicakirsh
⭐Jessica Kirsh on X: / jessica_kirsh
⭐Ryan Hansen Space on CZcams: / ryanhansenspace
⭐Ryan Hansen Space on X: / ryanhansenspace
⭐Ryan Hansen Space on Patreon: / ryanhansenspace
⭐Starship 3D on X: / dstarship3
⭐TDSN on X: / tdsn19
⭐ChromeKiwi on X: / ashleykillip
⭐Alex Delderfield on X: / alex_adedge
⭐LabPadre on X: / labpadre
⭐LabPadre on CZcams: / labpadre
⭐VirtualSpace_3D on X: @Lolomatico3d
⭐VirtualSpace_3D on Patreon: / vs_3d
⭐Nick on X: / chameleoncir
⭐Jerry Pike on X: / jerrypikephoto
⭐Tony Bela on X: / infographictony
⭐Ryan Hansen Space on CZcams: / ryanhansenspace
⭐Ryan Hansen Space on X: / ryanhansenspace
⭐Virtual Space_3D on X: / lolomatico3d
⭐The Ring Watchers on X: @RingWatchers
⭐Lunar Caveman on X: / lunarcaveman
⭐The Space Engineer on CZcams: / thespaceengineer
⭐Starship Gazer on CZcams: / starshipgazer
⭐Starship Gazer on Patreon: / starshipgazer
⭐Starship Gazer on X: / starshipgazer
⭐Greg Scott on X: / gregscott_photo
⭐www.nasaspaceflight.com
⭐Nick Henning on X: / nickhenning3d
⭐ErcX on Twitter: / ercxspace
⭐Alex Svan on X: / alexsvanart
⭐Caspar Stanley on X: / caspar_stanley
⭐Neopork on X: / neopork85
⭐Corey on X: / c_bass3d
📄Links for this Episode:
www.spacex.com
www.spacex.com/starship - Věda a technologie
Secure your privacy with Surfshark! Enter coupon code FELIX for an extra 3 months free at surfshark.deals/felix
What do you think? Are the Chopstick vibrations a serious concern? Let me know your opinion in the comments!
the attempt will not succeed. The arm need to be in the shape of the booster round shape.
Too much publicity/product/self publicity placement, I am getting tired of it and I have been watching since years (feedback comment)
If the hold downs change was just geometric, I would assume you would check only one....
they make it entirely non-viable, which puts them in good company with the rest of the misguided program
i find it hilarious that you are speaking about this actually working when the last booster came in backwards and on fire they are so far from reliably catching a moving object without 1 out of 2 hitting the tower its untrue
“One ping only” The hunt for Red October… awesome movie
In 1990 I was stationed on the submarine tender USS Orion AS-18 in La Maddalena, Sardinia. One night after The Hunt for Red October had premiered, the crew was watching the movie aboard ship and to top it off one of the submarines we were servicing just happened to be the USS Dallas. Talk about great timing.
@@user-ed7ve1tw6y that is awesome!
So damn true =)
@@user-ed7ve1tw6y Damn, this is something to keep as a memory =)
I think the booster should do a hop to practice the catch, if it works great, lot's of data and they can repeat the maneuver to build confidence ahead of the real return to launchpad catch. If it goes wrong then at least there isn't a whole lot of additional velocity coming down on the launchpad and maybe they can crash somewhere designated. Either way I think a hop would be a good move.
Talk about risk someone is gonna go up in a Boeing starliner lol.
Long as they don't do any whistleblowing before they'll probably be alright.
If NASA says it's ready for crew, it's ready for crew.
They have done multiple uncrewed tests, simulations and hardware upgrades to eliminate as much risk as possible, and now (after almost a decade of additional work) they are finally ready for a crewed flight test with minimal risk.
2 of them actually. What goes up...
@@plainText384
DEI decision.
@@plainText384 all of this might be totally cool and it might be totally fine. I love space and would love more then anything to go to space. But no way in hell would I get on that thing.
The vibrations look to me (I could be very wrong) like a classic robot arm control problem. The arm is sent to the intended position but overshoots and has to come back, but overshoots.... What you want the arm to do is smoothly approach the intended position and come to a stop at that position without overshooting. A similar problem occurs when a tank is traveling over uneven terrain and has to keep its main gun accurately aimed at a target. The solution is fairly well known and involves using laplace equation or even a simplification thereof to apply the correct 'smoothing' of the robot arm. In short it looks like a software problem to me rather than a mechanical one, and so should be relatively simples to fix.
Thinking about it, the vibrations look exactley like what I would expect from a system under development. You need the frequency of the oscillation in order to cancel it out efficiently. Get it wrong in the other 'direction' and the arm will never quite reach the intended position.
I'm impressed that you know about PID controllers. but the chop stick controller needs something a bit more sophisticated. (PID is just a quick COTS solution for simple feedback systems, e.g. some sort of fluid level control in an industrial process)
The chop sticks need to control position, velocity, and acceleration with accuracy. So they need at least a Type 3 controller. That means at least three integrators in series. But the hydraulic cylinder is an integrator (it's length is proportional to the accumulation of fluid inside it) so they only need two integrators implemented in hardware or software.
Important note: The ringing we saw when they moved the arm quickly and stopped abruptly was probably INTENTIONAL. They could be trying to verity their model of the arm's transfer function. In other words, they are probably checking it's vibrational modes.
You make it sound so simple, just 1,2,3 and all done.
@@Mentaculus42 PID is 123 !
@@marinmitu995
Maybe so, but what sx needs to deal with the control issues of the chopsticks is more complicated. But sx does have some competent control systems engineers but it is unlikely that they can band-aid a solution onto the existing chopsticks to get the desired results.
Nah, if you wanted to do it properly you would be adding in force feedback and other sensors, particularly one for detecting the catch and adding the manually calculated mass of the rocket. You could bake all this into a series of integrators (you would need MANY to account for the many oscillation modes and corrections) but you might as well just run a proper sensor-driven simulation since that is just generally easier these days and is way more flexible. I'd be surprised if they are not doing that.
On the whole I think that arm catch method is very silly though. It's perfectly doable but it requires a totally unnecessary amount of precision and opens up many failure paths (that they don't seem to be good at predicting). Arms should just be there for stabilisation and rely on a pad for vertical force. I do at least like the ability to move downward to compensate for some landing speed error, but that can be done far better with a moving pad since it just needs a huge vertical actuator that could be easily built to handle any crash that crew could survive (drop from 20m or such). All separate from the launch pad of course, even if it's just on the other side of the tower.
@@samheasmanwhite
That sounds interesting, someone who is thinking about it from a systems perspective. Maybe sx can borrow some tesla people that are working on FSD (F..l Self Driving) and that humanoid robot and graft that technology into a giant catcher for the rockets landings.
It seems that sx is extremely motivated to capture from the top to keep the rocket in tension to avoid buckling. Probably one of the reasons to not first start with a rocket that is both slender and mega big before working the “kinks” out of your recovery system. Definitely requires too much precision in too little time with too many failure paths while it is burning remaining fuel too fast with too little fuel margin all while trying to index & 3D position the huge rocket vs the oscillating chopsticks. Not sure that silly quite covers it considering that it all has to work flawlessly 100% of the time or things too bad to contemplate will happen to your stage zero. This should be entertaining to watch.
Tower 2 has short arms ?well it's kind of obvious it's going to be called T-Rex. I'm calling it- catchZilla and T-Rex
I was thinking alligator arms, but I like T-Rex.
T-Rex? Yes!!!!
Rexzilla?
With raptor engines
Instead of a landing tower, why not have a landing pit?
It's a big cylindrical hole in the ground that's a little deeper than the height of the booster. The diameter can be wide enough so that there is a large target for the booster, which would reduce the likelihood of missing. The booster is caught by a circular aperture door at the top of the pit that closes in toward the center catching the booster as in descends. The door can be designed so that the central catching aperture can be offset to the side a bit depending on where the booster enters as the booster would not always be perfectly centered. In other words there would be built in adaptability.
If the engines fail when landing, the booster explodes underground, minimizing damage to any surrounding infrastructure above ground. Even better, you can have an emergency carbon dioxide release system in the pit to smother the engines and prevent or greatly limit combustion thus reducing the explosive power of a hard landing. This system can activate once it's known that the landing engine isn't working as it should.
It's also even possible to design some sort of cushioning system at the base of the pit to soften a hard landing - maybe with water or some kind of net or something.
Digging a deep hole in wet sand is harder than building a tower. It can and has been done for subways etc, but it takes forever.
@@phik It might be harder, but it would probably be better in the long run.
No, it makes it unnecessarily complex. You would need to deal with the engine exhausts. Thus needing to calculate how big you would need to dig offshoot tunnels for that, and how to build them for them not to disintegrate after a few uses. And you still need to maintain them. Much more costly over time. Especially considering that not every landing might be perfect causing damage to any of the tunnels or the pit.
@@phik Consult The Boring Company for information.
I think a circular tower with a camera shutter type catching mechanism (or multiple ones) would be easier to construct than a vertical pit. If the mechanism could be perfected, the launch pad could be fully integrated too.
The arm shaking was most definitely part of the test and measurement process, to see what vibration mode and its amplitude would be.
Those white square patches on the chopsticks, "could" be preliminary placement of a set of Load cell sensors on the chopstick arms. For measurements of forces during the catch or even during the stacking. You would like those forces nicely distributed and within the design limits.
The counter movement would turn the wiggle into a wobble (because the actuators are at the hinge point but the mass is further out). Probably still too much motion for the precision maneuver of a catch.
You forgot to mention the SR75 launch in Australia that took place early this morning. Of course, that launch was hardly covered, and it seemed very low profile, but it is still a rocket going to space
The whole thing is insane, it is like trying to catch a flying California Redwood tree right out of thin air
I am so excited for IFT-4!!!! I feel it coming around the corner as everything is almost ready for it!!!
Yes, the T-Rex move was iconic, but the headphone bopping was hysterical.
I would like to see them fast track the fabrication of the second launch tower so it is capable of catching the booster on the launch after this one.
Mitigates risk of damage to their launch infrastructure.
They don't need launch capability from that tower right now so the initial fabrication could be simplified and streamlined.
When booster catches on the second tower work they could move the booster catch to the main tower and starship catches to the second tower.
Keep the higher risk operations further from the launch infrastructure.
Everything at Spacex is fast tracked
FYI, The crane is called Marvin because Elon loves the science fiction novel 'The Hitchhikers Guide to the Galaxy', and the dull robot is called Marvin.
The shorter arms means less mass so the arms can respond more quickly. The whole system has some springiness. But the unsprung mass is an unavoidable mass that will contribute to stresses on impact. Since the mass has to move on impact the lower the mass the lower the stresses on impact. Also the lower the delta-V the lower the stresses. Matching the speed of the arm with what it is catching reduces the delta-V which reduces stress. With all the testing SpaceX does I'm surprised they haven't unit tested this. Like, have a booster take off for about a few meters off the arms and catching it.
You guys are a total class act. Your #1 of 2 sources of information about space that I trust.
The ‘you will probably be forgotten’ for booster 1060 got me. 😂
Yeah, vibration in such heavy and large arms will be important to catch the heavy load.
Only nowadays, we are able to potentially cancel such things out at this scale.
Amazing! 😎
2:56 where i worked, we call this « transient response management » (but my translation might be not perfect, cause I'm just a French guy ^^).
Great Episode! Thank You Felix and WAI Team.
Thank you so much for the help, Owe!!! :)
Kudos, Kaplak to RFA!!!!
YOU GO BOI'S AND GIRLS!!!!
i think using the gridfins is still a better idea than those little pins, even if they can't hold it at half of their length, they should be able to manage at a quarter, and if they do have a suboptimal catch, they can quickly put it on the ground instead of having no grip, causing it to swing. i'm sure they've done the math, just makes me uneasy seeing the pins so small
The problem with the gridfins is having all the force on the motors would completely wreck them.
Those pins, they can make them bigger if they need to. Gridfins, would not be able to survive one landing.
Another idea is that instead of chopsticks the tower simply has a landing pad on the chopstick arms. The pad starts at a near matching speed of the booster, lowering downwards at an increasingly resistant speed, an opposing force until the booster has slowed down and is on the ground. The resistance controlled by AI. A pad with clamps?
PID has been a viable option for large hydraulic systems fro a long time. I used them with ships autopilot and fin stabilizers for years. Fin stabilizers are large, like sterships fins. Accelerometers come into plat as well. This is a solvable problem.
"One ping only" .. I cracked up!
@whataboutit. Are you able to give any information on commercial launches in Australia?
I have had to deal with issues that require input shaping in oil pumps. It can be a giant pain in the a**. Very small things like oil temp can change viscosity of the oil and hence the feed back pressures. In turn this can change your resonance causing either an over correction or under correction. So they need would need some way to adjust that in real time.
Consider catching a modified Falcon 9 booster that is built out to the specs of a regular booster to see the coordination of the catch and the tower.
One might use one of the older boosters launched to 10,000 feet and then landed in the tower. That would prove the catch and the older booster if lost would not be as big a loss.
Thank you for answering the "Landing tower" question
Great job Felix! So nice to have you back! You Rock!
Now it makes sense to make an catch tower you upgrade to an second launch one.
The sound wave canceling method is what is used in our quite helicopters.
Excellent visual on the PID 👏
Thank you for a very honest evaluation
Thank you for covering the upcoming launch by RFA
Nice, 420k subs guys! Smoke to your success!
3:30 its like a giant tuning fork! You ring one and the other vibrates. They should make one longer than the other and the vibration should go away.
I like pid tuning. I use it on my drone.
I think SpaceX should test with a scaled down version of the booster
Thanks!
A question for those who know more about SpaceX launches than me:
In the future when Starship launches become more and more routine, will they always have to conduct spin prime tests and static fires before every launch? Or will Raptor and Starship be able to get to a point it’s so reliable those tests aren’t needed?
No one can answer that. Raptor was designed and is being developed to be as reliable or more than commercial jet engines, and according to a majority of experts inside and outside Spacex theoretically there is nothing preventing that.
Probably ‘yes’, it’s sensible to always have those tests before launch on Earth.
Especially this early into the program.
Love your educational and fun videos even though I barely understand much of this. Thanks!
didn´t know that im a space x crane xD nice video felix enjoy it every time! keep this great work ^^
I’m so excited for ITF-4 and really hope SpaceX succeeds in catching the booster!
Definitely should have or should build a "catch only" tower, using the launch towers as backup. The risk of destroying the launch complex is very high. It will always be needed; the risk never goes away. No chance it will ever be obsolete.
Pretty cool episode !
Not having landing legs saves weight and catching the booster at its top means it will not be subject to being blown over by wind gusts like it might if standing on legs. The booster will be like an empty soda can in the wind after landing with little fuel on board.
Then dont fly when strong Winds occur...Problem solved. One failed Catch attemt and starbase is history. And so will the FAA Approval be. it will set SpaceX back Years.
You are forgetting the weight of engines located far away from catching arms. The arms should be flexible to dampen swinging of the booster.
Just enough fuel to blow up the entire area. Don't you remember those failed Starship landings? The fuel tanks were almost empty but still blew the area sky high.
Not having landing legs also means that a lot of complicated processes all have to function with great precision with zero margin of error and catastrophic consequences for failure. It's a bad trade off to save weight.
Yeah you’re exactly right smoothing out the acceleration/deceleration curve should solve excessive oscillation. It seems to start because of its immediate stop at its end of travel which then reverberates around the structure
PID controllers are 1950s technology that were originally implemented using analog instrumentation. Given the relatively slow speed of the chopstick arms, more sophisticated digital algorithms such as model predictive control could be used. This would get them up to 1980s technology.
@4:54 The problem with this _counter-movement strategy_ is that while this might be fine for the _chopsticks_ at the top, it will accentuate the motion around the engine nozzles, which will swing wildly by up to a meter.
At that point, I'd worry about the _chopsticks_ ability to hold onto the booster. (It _might, maybe_ work properly for the *StarShip...* )
This is pre-catch to allow the catch. Post-catch, a similar routine should be engaged to dampen the combined arm/vessel system.
Excellent info, video, graphics, editing, as usual! WAI delivers a top-notch production. And special thanks Felix for that unexpected 'short arm' visual demonstration....literal LOL 🤣
I would suggest a design similar to a gyroscopic boat damper to stabilize the arms
Ahhh were talking about control systems today.
Love the arms through the sleeves, a real LOL!
Maybe they are tuning PID with Ziegler-Nichols method. So they induce the oscillation to calculate proper initial PID components 🤔
DUDE... You started to talk about PID tuning... and I was thinking, OK, 3D printing next? THEN Input Shaping! You have my attention
PID tuning!
Now that is a skill I can appreciate!
A new Mechazilla with shorter arms will be more analogous of a T Rex at any rate.
It needs giant googly eyes. 😉
@@getinthespace7715 That would be hilarious 😂
@@Dinfes, Do you think if I took a trip to Texas, fabricated some 20 ft diameter googly eyes and dropped them off at starbase they would understand the assignment and weld them to the top of the tower. 😆 👀
I hope Elon sees this, I think it’s a great idea 🙌😂!
@@getinthespace7715 Make it where the rotating eyes are counterweights that dampen vibrations for the win.
I like your reminders of videos.
👍
shopsticks are oscillating because they work "empty", once they will have grabbed the starshing all oscillations will be cancelled by the opposite directions of oscillations (auto-cancelling) , the pinching force , and the weight of starship … i think …
as far as the records for launches, this is a lot like the record for recovery - they aren't even trying for that record, the record is incidental, not the goal. it's a demand for those launches.
Appreciate the concept all I can imagine is fumble sticks we all fall down. Too many variables one fail is reducing the economy of catch and reshoot to a net negative.
Maybe one tower is for launch and the other for catching hence the arm length difference
The engines wouldn't survive firing so close to the ground even if the mass of legs weren't too much. You'd need like 20 meter tall legs to avoid shockwave damage.
Where do you get that 20 meter number? The booster will be nearly empty and without a starship or payload on top. A few lighted engines with low throttle would probably be sufficient. Besides if 1 or 2 engines would get damaged they could be replaced at a fraction of the cost of the whole booster.
This news is pivotal to Spacex. Blue Origin dropped a video today about their launch site. I hope they finally have something WAI worthy. Let’s hope for coverage Felix!
I think that those new white pads on the Mechazilla arms are strain gauges not conduit mounts. This would provide valuable feedback to engineers about how much strain the arms are under when the booster lands in IFT5, and possibly data to feed in realtime to the catch algorithm when moving the arms.
Dig a hole in the ground and catch the rockets in the hole, at least for the proto-types and it will be safer
Motorcycles use dampers, ie. shocks or McPherson struts. Instant reaction and adjustable
Wie immer, super Video! LG aus Acapulco! Bin auch Deutscher, aber ausgewandert. :)
Chopstick wobble may require a bit more D to reduce the overshoot. Have to be careful with adding Derivative as you still want to hit your mark.
That cancel business works good until it amplifies the harmonics.
Seems to me that catch towers
and launch towers should be separate.
Propellant tank issues
Ich denke das SpaceX denkt ist es besser wenn der alte Tower kaputt geht als der neue. (Belastungstest) Der neue Tower kann dann noch wieder verbessert werden.
Space X could fit some stabilizing gyros of suitable sizes upscaling of the type used in small boats to control any shaking.
I think catching can definitely be done by SpaceX but the catch mechanism needs to be much higher and the arms twice as wide.
Like the landing legs of the Falcon 9, a pair of deployable short arms attached to the hard point and the stabilization booster`s point may be works to avoid the closest approximation and the roll orientation needed. If these are damaged they can be replaced beetwen flights
Felix I think ITF 6 will be a full orbital mission .. with a major concern on re entry catch n grab I fear the risk that i have what if it comes close to the catch n grab the booster might be off coarse by a fraction of a few feet and destroying the olm and the pad but I do understand the risk we just don't know what will happen we just have to wait and see. But I do not doubt the rapid success for spacex and the team.. I have faith in them good luck ITF4
No No No - not ready to catch that booster. In My Opinion.😨 And begin with separate landing tower.
similar to the problem of catching a moving cargo container to its waiting truck, up 60ft - across 50 feet - down 70 to within 1ft then jiggle, fighting the mass and rotation and pendulum forces to nail it every time, no input shaping works as the mass variable isnt fixed. no pid will be accurate as wind forces vary. years of container juggling has proven only one computer architecture has the right learning model to be 'on the money'...
With proper control system design the movement can be very precise. Wish this tech was applied to cranes and other machinery.
9:43 very realistic render! 🤣
Many thanks!
They should be more concerned with the boosters ability to do the last bit of the maneuver.
It was funny to see Marvin lifting such a comically small load!
Why has only one actuator on the chop sticks been replaced? Are they going to replace the other one before flight 4?
Are they gonna try to catch the booster only? And starship will land on imaginary landing in ocean? If that’s so I’d love to see video of BOTH landing attempts. Thank you FELIX for what you do. You too TJ Like your presentation also.
PID is very powerful to dampen movement
what about reaction control wheels? large ones with the appropriate weight to control unwanted oscillations
I’d like to see them more focused on orbital flight and spend more time in space than catching. 🇨🇦☀️😎
I think a stand-alone catch tower is the better solution and a catch tower on a drone barge might be the best solution.
An easy fix for the arm occilation is to use torsion bars to keep it always under tension. Like a vehicle suspension in some ways, don't let it be able to move freely.
Curius what would be the minimum concequences or delays after say a missed catch that hits on land? How much red tape will be thrown their way after?
I am wondering if a couple steel cables in parallel would be strong enough but still provide a way to disperse the energy of the starship on landing. I don't know, I was just considering a second possibility for catching starship if the chopsticks don't work out.
Your tyrannosaurus impersonation was hilarious... :D
The valve actuators on mechazilla were originally digital, either opening or closing with no speed control. They will need analog actuators, which allow speed control. There is no PID in a digital actuator.
You can do ‘time integration’.
@@peterclarke3020 You can't do anything with digital actuators other than open or close them at a fixed rate of speed. Though you can use snubbers to make the speed change at different phases of stroke. There are many ways to skin this cat.
Felix, with regard to the Shaking of the arms, it could be a few other things. First, any looseness in the mechanical system would have to be corrected. This could be a pin fit that is too loose, or something along those lines. Second, Making sure all the air is out of the hydraulic system is critical. Any air in the system would be like bouncing a ball on concrete, which is what you see when the arm stops. Third, a validation there is not an issue with the servo valve. Fourth, a slightly feathered ramp to stop can help, kind of like when you let off the brake when stopping your car so it doesn't jerk to a stop. This will certainly be solved, much larger equipment than this is regularly moved in industry without the shaking, it will just take some work and tuning. Shorter arms won't necessarily help if the problem is one of the issues I mentioned above.
2.58 Better a LQR control law
Linear quadratic Regulator.
I personally would place a catch tower much further away from launch. in fact I think the tower is to close to the fuel for my piece of mind.