Beyond Augustine
Illustration by Mark Maxwell/Skycorp
Dennis Wingo: The Augustine commission, in its last public meeting in Washington delivered a stunning blow to NASA in finding that the program of record, (the current architecture) was not affordable, giving the FY2010 run out of the NASA budget. At first, this was a cause for consternation for many of us, thinking that a repeat of 1993 and the retreat from exploration is on the way. However, I don't think that this has to be the case as the president seems to be supportive of exploration beyond low earth orbit, just not overly generous with the checkbook. This is ok, and maybe this dose of reality from the commission can begin a new thought process for space exploration. In fact, it may be that the Augustine commission, by being this honest about the course that we are on may finally lead to some progress in exploration.
Breaking The Highlander Syndrome
Several years ago a space advocate named Ed Wright coined the term "Highlander Syndrome" to denote a mindset to where "there is only one way to do space" (referencing a cult Si-Fi movie from the 80's). Now that the Augustine commission has basically said that we cannot afford to go beyond Low Earth Orbit through the Apollo/SEI/ESAS heavy lift paradigm, which has basically ruled the "one way" thought pattern of NASA as unaffordable, what do we do?
In the 1990's NASA had its Orbital Aggregation and Space Infrastructure System (OASIS), that was a good stab at a non heavy lift architecture for lunar exploration. Going even behind that, the idea from the Human Exploration Initiative 1989, 90 day study of using the space station as an anchor to a reusable cislunar human spaceflight transportation system. This makes enormous sense in that if you reuse a lander or a transfer vehicle only once, you save most of the estimated $500M dollar price tag of each of them.
We have several spacecraft today orbiting the Moon and taking unprecedented multispectral, optical, and radar data, probing the Moon for its secrets. Those of us in the lunar community have been waiting for 30 years for this quality of data. What if they find something? Hypothetically speaking, what would be the effect of finding a more water on the Moon than the estimates that most people use? Would there be enough water in the polar regions for propellant? What if there were? Just think of what could result if we had an emphasis on reusable spacecraft for transit to and from the Moon, and a Single Stage to Orbit (SSTO) no from the Earth, but from the Moon?
There are many ideas that have been scoffed at or sidelined, like In Situ Resource Utilization (ISRU) that could be brought back into the mix. Ion propulsion space tugs are also much closer to reality than what most people think. Advanced computers and electronics could be introduced into space as well. There is no reason for not flying state of the art computers, software, and advanced analysis capability. This reluctance to use the advances in computers in space that we take for granted on the ground is a major impediment to lowering the operational costs of space exploration.
I noticed a huge wedge in the Augustine commission power point charts that Dr. Sally Ride presented that represents ground operational costs. If you shifted the power of a desktop workstation to a reusable cislunar lander or transit vehicle, you could dramatically reduce ground ops costs. The Gene Kranz model of ground operations is only necessary when the local system is too primitive and dispersed. The Kranz model was completely appropriate for the Apollo era, but maybe not today. If you had, say a MacIntosh workstation running Labview, Matlab, and Satellite Tool Kit, three excellent software packages for this type of activity, you would have real time computer control tied to real time sensors to give you all the data that you would need to successfully navigate in cislunar space.
This one workstation represents more computational horsepower than the entire computer industry in the 1960's. Could you imagine the star ship enterprise being controlled from the ground? This would cause a shift in operational planning that would dramatically lower its costs and the total cost of space missions. Who cares if you have to wrap it with 50 kilos of lead, it is still a bargain.
There was a budget wedge presented, that if implemented, would make all the difference in the world. On page 33 of Dr. Ride's charts a reinvigorated technology program is planned and its wedge is in the budget. This is a good thing. What we have learned by the exercise in ESAS, as well as SEI and before, is that congress is not willing to fund the type of program that senior people wanted, so we now have to figure out what to do with the amount of money given. It is my strong personal opinion, after traveling to Europe to build a on orbit servicing satellite, working with NASA and the defense department on various technologies and systems, is that all of the parts that are needed to field advanced systems exists, they just need to be brought to operational status (TRL-9 in the lingo).
A robust technology development effort that funds ISRU research, ion propulsion, In situ food production, space computers and software, could help us leapfrog past the limitations of not having an Ares V class vehicle and put us into a position of not just flying a few with the right stuff, but to be able to fly anyone, for no more than the cost of an ISS ticket today. People talk about single stage to orbit vehicles like that is the only other solution. Today it costs about $10,000 per kg worst case to get mass into low earth orbit. It takes $80-100,000 to get that same kilogram on the lunar surface. Where would reusability make a greater payoff?
If we take what we have, a space station and some form of human spaceflight, and figure out how to integrate both of these systems, which according to all of the options we will have, into the plan it is possible that we could come up with an architecture that is affordable and would allow the human development of the Moon and beyond. Therefore the technology development wedge is probably the most important part of the new plan, and is common across all of them. Lets just hope that they get someone good to run that shop at NASA.
There is great hope today for a good outcome for space. Realizing that the path that we have tried to develop for thirty years is no longer viable, that is unless the president is willing to put forth a substantial increase in the NASA budget and many are advocating that. However, if we don't get that money, which history shows is the likely outcome, then we must have a way to do exploration within the budget that we have. I am strongly confident that this can be done and that the best is yet to come in space exploration and development. This is especially true with the strong hat tip in the Augustine commission toward commercial LEO human spaceflight.
Kudos to the Augustine commission for straight talk and crossing my fingers that NASA and commercial enterprise will be exploring together and that we do it with 21st century solutions, not a cast back to a bygone era.
Computers in space
Right now, the fastest rad-hard processor available is a 100Mhz RAD-750 from BAE. Cost: $750K with 36MB RAM. But it's rad-hard - VERY rad-hard. They're flying on both LRO and LCROSS and 2 months into the programs neither has had a hiccup yet - and they aren't expected to.
But that's the price you pay for bullet proof hardware. It's INSANELY expensive!!
SO: If you're willing to tolerate errant instructions and processor resets induced by radiation you can do it WAY cheaper. That's what Dennis is proposing and his ideas have ALOT of merit. But the software becomes VERY DIFFICULT when you start talking about single-string processors running non-fault tolerant processes like orbital insertion burns and other space-like things. Right now the software is almost BRAINDEAD SIMPLE in comparison to what would be needed for these kinds of autonomous systems. And the price you pay for simple flying systems is just what's described: huge costs for ground operations personnel. We humans have to carefully monitor every aspect of the on-orbit machines, and send up highly detailed sequences of instructions which are VERY carefully thought out and tested.
On the other hand, we have single-string spacecraft up there which have run flawlessly for years. You tend to hear about the MRO rebooting or one of the MERs faulting, but EO-1 has been flying for 8.7 YEARS without a reboot. The code is RAM is the code loaded from EEPROM when the spacecraft was booted up before launch - in November 2000! Put that in your Mac, Linux, or Windows machine and smoke it! :)
Additionally, there are processors cores (logic) out there today which we can put into an FPGA and run in parallel voting schemes to mitigate radiation effects. Rad-hard FPGAs have come a VERY long way in recent years and this may be a way to turn out higher-speed RISC architectures for much less cost and still be radiation TOLERANT instead of radiation HARDENED. It's VERY hard to do this, but it's been done before and once we got a triple-voting PowerPC into an FPGA we're there, and the second processor costs single-digit thousands instead of 100's of thousands. But as Dennis mentioned, getting these systems to TRL-7+ is hard and very costly.
But the cost of not doing it is probably much higher.
Emory Stagmer
Project Software Lead Engineer for NASA's LCROSS Program
Northrop Grumman Technical Services STS group
Speaking only, of course, only for my self and not NG or NASA.
Computers in space
You lost me when you started talking about computers for autonomous space probes. My understanding of computers in space environments is that they have to be radiation-hardened. This takes time, and, in the fast-moving world of semiconductors, means that you're starting with a CPU that's half the speed of a current model. Second, vehicles, either in space or on Earth, generally use embedded processors, which trade speed for lower power consumption. This is especially important in outer space, where power is expensive in terms of spacecraft mass. Finally, you have to deal with the unreasonably long development times for building a space probe. You can pick a top of the line radiation-hardened embedded CPU for your spacecraft today, but it'll take you five years to get the thing built and launched, and by that time your computational power will be a fraction of the contemporary desktop systems.
By the way, I remember years ago you were trying to get a Mac workstation launched into orbit. Were you planning to send it up with ECC RAM?
Computers in space
Matt,
The fact that by the time the vessel is done it won't be top of the line anymore is irrelevant if it can do what its suppose to at the start.
Modern radiation hardened CPUs are still decently powerful. A RAD750 still ought to be able to be competitive with what NASA had on the ground in the 60s.
_In theory_ you could even build a Mac out of RAD750 if you could get Apple and relevant ISVs to support it. (The "G3" Macs used IBM 750s. My iBook has an IBM 750.)
You'd need eleventy billion megs of RAM, but if its a reusable vessel, the cost of the computer _hardware_ isn't that bad even if you need a cluster of $200,000 RAD750 boards.