New York:
For $150 billion, the National Aeronautics and Space Administration(NASA) could have sent astronauts back to the Moon. The Obama administration judged that's too expensive, and in September, Congress agreed to cancel the program.
For a fraction of that -- less than $200 million, along with about $250 million for a rocket -- NASA engineers at the Johnson Space Center in Houston say they can safely send a humanoid robot to the Moon. And they say they could accomplish that in a thousand days.
The idea, known as Project M, is almost a guerrilla effort within NASA, cooked up a year ago by Stephen J Altemus, the chief engineer at Johnson. He tapped into discretionary money, pulled in engineers to work on it part time, and horse-traded with companies and other NASA units to undertake preliminary planning and tests. "We're doing impossible things with really very little, if any, money whatsoever," Mr Altemus said.
A humanoid dextrous robot -- at least the top half -- already exists: Robonaut 2, developed by NASA and General Motors, is packed on the shuttle Discovery, scheduled for liftoff on Wednesday.
Bound for the International Space Station, it will be the first humanoid robot in space. It is to help with housekeeping chores at the space station as NASA learns how astronauts and robots can work together. Eventually, an upgraded Robonaut is to take part in spacewalks.
Project M also draws on other NASA projects that were already under way, including rocket engines that burn liquid oxygen and methane -- a cheap and nontoxic fuel combination -- and an automated landing system that could avoid rocks, cliffs and other hazards.
Integrating the technologies into working prototypes sped up development. "That's the magic," Mr Altemus said. "A lot of times technologies end up in the lab cooking, and then there's this valley of death where they never get to maturation or to flight."
Project M's planners say that a robot walking on the Moon would capture the imagination of students, just as the Apollo Moon landings inspired a generation of scientists and engineers 40 years ago.
"I think that's going to light a few candles," said Neil Milburn, vice president of Armadillo Aerospace, a tiny Texas company working on Project M.
But as NASA's attention turns away from the Moon -- "We've been there before," President Obama declared in April -- the prospects for sending a robot there are at best uncertain.
The quandary over Project M encapsulates many of the continuing debates over the future of the space agency: What should NASA be told to do when there is not enough money to do everything? What is the best way to spur advances in space technologies? And given the costs and dangers, how important is it to send people into space at all?
"The tricky part is whether it fits in the agency's framework for exploration," Mr Altemus said.
Last year, a blue-ribbon panel was reviewing NASA's human spaceflight program, in particular an ambitious project called Constellation to send astronauts back to the Moon. Although NASA has spent $10 billion on Constellation, most of the program is to be canceled when Congress finishes work on the 2011 budget.
Mr Altemus, for one, was frustrated by criticism of NASA that emerged during the Constellation debate and elsewhere. "I always felt like our organization was a Ferrari, and we were never allowed to drive with our foot on the gas," he said. "We were kind of at idle speed all the time."
Talking to his son at his kitchen table, Mr Altemus wanted something that was exciting but not so big that it would require years of deliberation. The idea popped into his head: a walking robot on the Moon, one that could send back live video, in a thousand days.
Mr Altemus took it to his staff the next day, telling them, "Let's do something amazing."
He recalled: "I said, 'Will you get behind me if I put this into the organization? I don't know if we can do it. I don't know if we'll get the money for it or will get approved -- let's try.' And so we just started, and it caught like wildfire."
Sending a robot to the Moon is far easier than sending a person. For one, a robot does not need air or food. And there is no return trip.
The thousand-day deadline was arbitrary, said R Matthew Ondler, Project M's manager. "It creates this sense of urgency," he explained. "NASA is at its best when it has a short time to figure out things. You give us six or seven years to think about something, and we're not so good. Administrations change and priorities of the country change, and so it's hard to sustain things for that long."
For the purpose of aiding science education, a thousand days fit easily into the four years that a student spends in high school or college. By contrast, even if NASA achieved Mr. Obama's stated goal of sending astronauts to an asteroid by 2025, a 7-year-old today would have already graduated from college.
To get the parts they need, Mr. Altemus and Mr Ondler have resorted to barter. Boston Power gave them a $300,000 prototype of an advanced lithium battery in exchange for engineering help on battery management issues.
"It was an easy trade, so we made several deals like that," Mr Ondler said.
Armadillo provided a prototype it had built for a lunar lander competition, and NASA exchanged engine technology and access to test facilities.
NASA also paid Armadillo about $1 million, but NASA's traditional development processes would have cost more and taken longer. In six months, the lander flew 18 times under tether and twice in free flight.
Not all the flights went perfectly, which was the point. "It's OK to put a hole in the ground once in a while," Mr Ondler said. "It's OK to have flame coming out of the wrong end of the engine once in a while, as long as we're learning quickly and building and iterating."
Mr Ondler told the story of an engineer going to Home Depot to buy about $80 worth of materials to test whether fuel sloshing in the tanks could destabilize the lander during descent. "From that, we were able to confirm our math models and design the full-scale test," he said, all in two weeks.
Project M slipped under the radar of everyone else in NASA, including the administrator, Maj Gen Charles F Bolden Jr. In February, in response to a question about projects that NASA might undertake with other nations, General Bolden cited a two-legged robot that the Japanese space agency wants to send to the Moon by 2020.
"Do I think I can do that?" General Bolden said. "Probably not."
At that time, the Project M team was hoping to get a go-ahead to start in March and accomplish the robotic Moon landing by the end of 2012.
Despite the sophistication of the project, the robot's capabilities would be slight compared with what a human could do on the lunar surface. Project M was conceived as a technology demonstration, not a scientific mission.
One of the main tasks envisioned for the robot would be to simply pick up a rock and drop it, as part of an education program broadcast to schools. Students could do the same and compare the relative gravity of Earth.
Work continues on Project M, which has cost about $9 million so far. Armadillo is building a second prototype lander, but there is no money for other aspects, like finishing the legs for Robonaut. Mr Obama's vision for NASA called for investing $16 billion over five years for space technologies, but the compromise blueprint drawn up by Congress shifts most of the money to a heavy-lift rocket.
The project did spark interest among the International Space Station managers, which is why a Robonaut is heading there. "I'm excited to see how we can evolve the technology in space and actually have a pair of hands and a working humanoid dextrous robot on the space station," Mr Altemus said. "It's a big move forward for the agency."
But for now, the plans for sending one to the Moon are on the back burner.