7 ms·
Rocket science is easy when you have a very powerful general optimization package available. Classic rocket science is trying to find closed-form minimum-fuel s
by Animats 6y ago
Rocket science is easy when you have a very powerful general optimization package available. Classic rocket science is trying to find closed-form minimum-fuel solutions for those problems. See "Hohmann orbit" and related topics. That's hard.
Legged running for robots is a very similar problem. It's under-actuated (more degrees of freedom than actuators) and constraint-limited. I worked out some closed-form solutions for legged running with anti-slip control back in the 1990s, which were later used in robotics work at McGill. Now you can probably just crunch a solution out of an optimizer.
(I used to make the comment about that, "It's rocket science. Which is good, because rocket science is a solved problem.")
- bpodgursky 6y agoI get where you are coming from, but if it was truly "easy", there would be a lot more self-landing rockets. It makes one almost-impossible thing... possible. Not "easy".
- FriedrichN 6y agoIt makes me think of the Laputans in Gulliver's Travels, the race of scientists that are so caught up in their theories they can't do anything practical.
- knuthsat 6y agoI mean, if you have perfect measurements of the state of the rocket and a complete model of it then it's just physics. Problem is that measurements are random variables and your model of the rocket is probably not perfect and now somehow you need to make it work.
- colordrops 6y agoEverything is "just physics"
- knuthsat 6y agoThat's true, I guess I was imagining more of a Newtonian approach, modern physics has everything.
- ClumsyPilot 6y agoNewtonian model gets you a rocket in a vacuum without atmosphere or vibrations made of indestructivle materials. In the real world there is the atmosphere, fluid dynamics of the fuel, combustion stability, material science as engines have to withstand hot and chemically aggressive exhaust, etc.
- mschuster91 6y agoNo matter how difficult the problems are - it still costs billions of dollars to develop a rocket program, and disposable rocket programs have the "advantage" for the industry of a steady order supply... which means the industry has no incentive to push for reusable rockets. SpaceX has the advantage of being funded by a billionaire who isn't beholden to any political demands (including job creation/upkeep), and so they can do whatever promises the biggest profit.
- ctdonath 6y ago“Solved” in the way that a mathematician views a bucket of water and a hotel room fire as “solved”.
- inopinatus 6y agoWell, of course; having demonstrated the existence of at least one fire brigade, actually containing a blaze is best left as an exercise for your graduate students.
- codeduck 6y agoShots fired.
- perl4ever 6y ago>closed-form minimum-fuel solutions My dad used something called variational calculus to solve physics problems. It's beyond my understanding, but I gather it was more useful in the slide rule days. A former co-worker of his commented that once you could just throw computing power at a problem, being able to construct the elegant solutions was basically obsolete.
- monocasa 6y agoYeah, there was a similar comment from one of my greybeard coworkers. That one of the biggest changes he had to get over as engineering moved on over the years was the gradual shift from very symbolic, human friendly (one might say 'elegant') mathematics into more brute force, pure crunching techniques.
- ur-whale 6y ago> was the gradual shift from very symbolic, human friendly (one might say 'elegant') mathematics into more brute force, pure crunching techniques. Conversely, in these days when anyone can go buy themselves a couple of teraflops at the shop round the corner, when you look at the way they solved problems back in day it feels very strange. But then, they didn't have a choice of course. However, when you look at modern physics courses (worst offender IMO is standard quantum mechanics courses), where huge swaths of times are expanded trying to find closed form solutions to Schrödinger's equation, it feels very bizarre (and a giant waste of time).
- selimthegrim 6y agoJust the other day I used variational techniques to show the bound on a Mott transition in terms of electron density to an undergrad who had been complaining to me that they "hadn't learned any of this stuff in class just Dirac notation"
- bodhiandphysics 6y agoTrust me, calculus of variations is still used! It’s even more important now that we have lots of computing power. Actually, the major way we solve differential equations, the finite element method, is a variational technique!
- FriedrichN 6y ago> because rocket science is a solved problem Some time ago I wondered why rockets keep blowing up because I thought rockets would be a solved problem by now. Given the time and money involved in launching an object into space there is every incentive to make it go right. I'm not an expert in rocket science, physics, or mathematics but seeing rockets blow up before having left the atmosphere makes me think it is not a solved problem.
- hypertele-Xii 6y agoYou're thinking of engineering, not science. The math is more or less all figured out. But how to apply it so things don't break? That's engineering.
- FriedrichN 6y agoSemantically that's correct, but when talking about rockets people think about actual real rockets, not theoretical ones. I guess this is one of those instances where the words we use and their exact meanings differ from what we mean.
- ClumsyPilot 6y agoEven if you are going to advocate that view, there is more science to rockets than just the rocket equation. From combustion instability, to materials and design of closed cycle engines to behaviour of materials in space it touches on many areas of science where the answer can not be off by 3% or the thing explodes.
- vimacs2 6y agoAbsolutely correct. Arguing that "rocket science is a solved problem" just because we have the rocket equation and the rest is just engineering is basically saying that condensed matter physics is just engineering because we already have a deep understanding of the behaviour of elementary particles in isolation. Reductionism is a useful conceptual tool but it's important to realise that's what it is, not some sort of ultimate arbiter of truth. You need to consider a system in it's whole as well as in it's parts if you want to truly grok it.