How would you like to do your next project at one-sixth the price and in half the time, with a mandate to offer scores of capabilities never before available, and have one shot to get it right? Did I mention that most people on the planet would be watching?

If you're used to similar demands, you might think that such a project "isn't rocket science." But in this case it was. Brian Muirhead, chief engineer at NASA's Mars Science Laboratory, continues to share what he learned as project manager for the 1997 Pathfinder Mission to Mars, developed and managed by NASA's Jet Propulsion Laboratory.

As a keynote speaker at NI Week 2003, Muirhead asked, "How would you like to be in the position where your CEO asks you to 'Take risks on this next project, but don't fail'? I think most people would be sprucing up their resumes." Failures came along the way, but the mission contained the right balance of ingredients to ensure overall goals were met. How?

• Test, test, and test. Wrapping the payload in 19-ft airbags allowed use of parachutes and much smaller retrorockets for a bounce-down landing.
• Co-locate and communicate. Putting hardware and software teams together was a first. When a programmer was trying to control a real-time function, he quickly got a hardware trigger after making an "over the wall" request.
• New technologies developed, focused, and proven in the heat of a project can be essential to overall success. Thousands of tests and modifications during design allowed in-mission corrections, such as untangling the rover from the airbags.
• Develop robust solutions tolerant to "unknown unknowns." Adding small rockets to help slow the lander was counter to the simple-as-possible design, but added necessary safety margin.
• Taking risks without failure requires license to do things differently. For the first time, NASA didn't write a significant amount of software. Instead, a RSC-based flight computer on VMEbus was used with Wind River's VxWorks operating system and National Instruments' LabView for rover navigation. Landing within 22 km of target was on the same scale as hitting a hole in one in Austin, TX, from Los Angeles, Muirhead says.

The Pathfinder team found creative ways to design, test, fix, and do it again. That sounds like what control engineers do daily.

Mark T. Hoske, Editor-in-Chief, mhoske@reedbusiness.com

Online extra for September 2003 ‘Think Again: Down to earth and a world away’

Pathfinder Mission: Added tips and facts

Here are additional tips learned and facts about the 1997 Pathfinder Mission to Mars. Brian Muirhead, project manager for that mission, is now chief engineer at NASA’s Mars Science Laboratory. He spoke at NI Week 2003, an Austin, TX, conference in August for National Instruments customers, partners, staff, and students.

Additionally, Muirhead says, taking risks and avoiding failure requires:

Brian Muirhead, chief engineer at NASA’s Mars Science Laboratory. Photo by Mark T. Hoske.

• That the team be built on talent, trust, and commitment.
• Hands-on team leaders should have essential micro-knowledge (not micromanagement) for fast, effective decision-making.
• When in the faster, better, cheaper mode, don’t compromise the quality part of “better.” While Mars Pathfinder was NASA’s “faster, cheaper, better,” a sister “polar lander” mission probably also would have succeeded had the team not over-squeezed on cheaper—a missed test might have corrected the point of failure.

In comparison, the Mariner mission in 1962 cost $122 million, the Viking mission in 1976 cost $1.16 billion. Pathfinder’s goal was to land on Mars, with half the preparation time (just three years), and include a rover with onboard science, for $150 million (later revised to $171 million, significantly less that about $3 billion spent on the previous NASA mission). In the end, the Pathfinder managed to return to NASA $10,000 of $171 million.

Pathfinder took off on Sept. 4, 1996, and landed safely on Mars July 4, 1997. To bleed the energy from an 18,000 mph approach within five minutes required 44 events to happen in sequence. No one previously incorporated airbags for the bounce-down landing. Testing of components and end-to-end testing with multiple computers of the sequence “brought a Cray computer to its knees.”

Signal time from Mars to Earth ranges from 7 to 20 minutes, depending on orbits of each. With that delay, real-time control is impossible, so the lander and rover have some on-board intelligence (about as smart as a bug, Muirhead says). The rover used computer-enhanced remote driving, where the operator picked points to drive to. The rover would stop and phone home when in trouble.

Plan was to create about 20,000 lines of code; 150,000 lines of code launched. Typically, Muirhead said, that would cost 800% more, but related costs increased about 15-20%. The lower-than-expected cost was because the seven people on the programming team were systems people first, and darn clever coders, secondly, he noted.

Lander and rover returned enormous amounts of data and images of the surface, atmosphere and weather from Mars. The rover, designed to operate a week, worked for 10 weeks; the lander, designed for a month, operated for three months, Muirhead said. Information gathered has been critical for subsequent missions and for fueling the imaginations and dreams of future engineers and other supporters of space exploration.

The Mars lander planned for launch in 2009 includes a rover that weighs almost one ton (compared to the microwave-oven-sized Pathfinder rover), which, Muirhead says will be a cross between an SUV and a Mazda Miata. It will spend a full year on Mars.

Among NASA sayings is that exploration and discovery are, as they have always been, humanity’s pathway to the future. Spread the word.

Here are links for more information:

NASA
National Instruments
Motorola Computer Group
VITA (VMEbus organization)
Wind River