• Content count

  • Joined

  • Last visited

Community Reputation

214 Excellent

About RuBisCO

  • Rank
    Rocket Scientist
  1. Put wings on it and add under-structure, hence the higher structure to fuel ratio I used, which is typically 5%, not 10%-30%.
  2. Again I'm not talking about a plane, just a rocket, a rocket with wings to glide back to a runway once it has spent its fuel. SpaceX may do even better by replacing the mass and drag of wings with rocket fuel for a controlled vertical landing. So the need to fly through the air on fuel efficient jet engines is non-existent.
  3. Nope, I never said such a thing, only that fuel cost are not the concern, $30 a kg clearly would mean the fuel cost is going to be far lower then simply the maintenance and launching costs, regardless of the fact is consumes a lot of fuel. They were hoping for launch costs with a space shuttle in adjusted dollars of ~$1000 per kg, so the fuel costs would be 0.3% of that cost. There is clearly no need to make a more fuel efficient spacecraft for getting to LEO. What matters more is turn around time, maintenance and launch costs.
  4. Look at the weight of the space shuttle, ~25 tons of cargo to orbit with a on pad weight of 2,030 tons. No, my proposal is simply a fully reusable space shuttle, as it had been originally proposed, two winged stages would take-off, first stage would glide back and second stage would enter orbit, drop off cargo and glide back. My belief is had they actually built that, instead of the semi-reusuable POS they created, the space-shuttle would not have been the utter failure it had become. As for the fuel use, according to Elon Musk the fueling cost for the Falcon 9 is $200,000, I'm hypothesizing a rocket that uses 3 times as much propellant, so that is $600,000 for the first stage in propellant. The second stage propellant cost using $3.66 per kg of LH2 and $0.16 per kg of LO2 and a O2:H2 mass ratio of 6 = $151,722 So total costs in fuel would not exceed $1 million, for 25 tons to orbit that is a cost of $30 per kg. I think that focusing on the more efficient design would require soo much development dollars upfront it simply would not be worth a much simpler design using older technology.
  5. Here are the rough calculations I get for a two stage to orbit space plane, first stage Kereosene-LOX, second stage LH-LOX. First Stage Second Stage Fuel 1500 200 t Ratio 10% 30% Cargo 285 25 t Wet 1935 285 t Dry 435 85 t ISP 320 450 s Delta V 4684 5339 m/s Total 10023 m/s Prop. Density 1 0.28 t/m^3 Volume 1500 714 m^3 Tank D 5 5 m Tank R 2.5 2.5 m Length 45 25 m # of Tanks 2 2 Volume 17012 916 m^3 Vol. Ratio 113% 128% Looks very doable, would not require atmosphere compensating engines, a structure to mass ratio of 10% is realistic for a kerosene-LOX space plane. First stage would consist of two 5 m wide tanks strapped together (side by side) that are 45 m long, second stage would consist of two 5 m wide tanks 25 m long. First stage may be able to glide back or use jet engines and residual fuel to fly back, second stage would enter orbit and glide back.
  6. Ok, another question: can the AI use the rocket pods?
  7. Yes the firing backwards things... what I mean is for example the target is north, the plane (with BD pilot on) does some maneuvers to avoid incoming missiles from the target, between missile volleys the plane will attempt to fire on the target, but it won't necessarily line up with the target (face north) it will even launch mavericks south sometimes (180 degrees from the target). I think this happens only with the FLIR ball as it can lock on to a target behind the plane. The FLIR pod can't look backwards and I have not seen it try to launch maverick missiles at such an extreme angle of attack. The only problem is the pod takes up one missile rack.
  8. Well is there any way of turning off predicting which one has a better chance to hit?
  9. Just replace its AA missiles with AG guided missiles and bombs.
  10. Yeah I tried changing the order I put them on the craft, no change. Also with the FLIR ball sometimes it will lunch missiles backwards, which is a waste. Any way to tell it to launch maverick when the target is in front?
  11. I'm having a problem with my attack aircraft, it tends to discharge missiles from left wing first, making the plane unbalanced. Is there a way to get BD armory to use missiles/bombs in a specific order?
  12. Thank you I now have something to listen to on my commute.
  13. Well yes evolving programs are software, but I'm also talking about HARDWARE artificial neurons on a silicon chip undertaking the same kind of operations as our brains electrically, such as to achieve Strong AI, that can self repair and maintain a civilization of immortal sentients for billions of years, and thus forgoing the need for frozen corpsicles.
  14. When we have a program on a computer, errors do occur as we copy that program, generally these leads to the program not working at all, but in theory a error that leads to a new operation as possible. In fact there are evolving programs that have specific sections of code that random numbers are throw into to produce mutations, the mutants that work better are selected for, breed and repeat. You are complete hung up on this terminology. a neural network and change its pathways and learn completely new things, it is not a program. The nature of how it learns it identical to how organism learn and can even have thousands of separate networks learn, compete against each other, kill off the losers, breed the winners and repeat for generations, in virtual space, in just hours. Take some time to learn about neuromorphics: That because the AlphaGo program is a simulation of a few million neurons (less then a frogs brain) and it was raise or breed to do nothing but play Go. What will happens when we take hardware neural networks of over 80 billion neurons and 80 billion x >10,000 synapses and dendrites firing at the speed of electricity and not at the speed of ionic gradient waves (firing speed will be at the millions of Hz verse our neurons that top out at 1000 hz), we take that and raise it up to like a child? No that is exactly what evolving programs do. We have gone from 256 artificial neurons on a chip 5 years ago to 16 million today, we will be at over a trillion neurons by the end of the next decade.
  15. Yeah, no your wrong Everything mutates, all data corrupts over time is only a matter of rate of mutation. As for this false dichotomy of "program to do" verse AI, no that is totally incorrect. We learn via a neural network in which neural connections are varied at random, pathways that produce a desired outcome are strengthened, pathways that don't are downgraded and eliminated, a process of evolution scaled down from individual organism to synapse strength. Mind you we also learn though conventional evolved neural componants, we evolved to mimic, to learn language and process data in symbols, etc. All of this can and is being emulated! Consider the latest competition between the worlds champion Go player and AlphaGO AI. AlphaGo was not programed to play Go, rather it consists of a simulated neural network, that was shown thousands of go games, then played its self thousands of times with different variants, the winning variants were bred together and mutated and repeated, evolving AlphaGo to levels where it crushes the human players. So in conclusion: yes machines can mutate (we are machines at a fundamental level in fact) yes machines can evolve (at the very least en silico) and machines can learn as we do with neural networks.