Interview: The driver behind NASA's Mars Rovers
- 16 November, 2007 11:15
Behind every robot is a driver. While NASA's twin robot geologists, Spirit and Opportunity, have gained plenty of media attention since they first landed on Mars at the start of 2004, little attention has been given to the team of dedicated IT workers behind the robots, plotting their every move.
We talk to Scott Maxwell, one of 14 Rover Drivers that work in NASA's California-based Jet Propulsion Laboratory, and find out what it is like to control a robot on another planet. Maxwell discusses what makes up an average work day, the highlights of the project, how he got the job, and the tools he uses in his work. Ashley Stroupe also discusses her life as a robot driver in the next part of this two part series.
What in your opinion are the top three discoveries the Rovers have made?
The Rovers went to Mars to help us tell the past story of life in the solar system by looking for evidence of past water activity on the Martian surface. Both Rovers have found such evidence, so I'd choose two of the top three discoveries as independent confirmations of past water activity on Mars -- Spirit on her side of Mars, in the Gusev Crater, and Opportunity on the other side, in the Meridiani Planum.
But maybe their most significant accomplishment has simply been proving that this can be done. Before Spirit and Opportunity, we'd never successfully operated free-roaming Rovers on the surface of another world. It was, in part, an act of faith that we could do it and that it would produce good science. But both Rovers have shown, quite conclusively, that this is not only a terrific approach from a science point of view, but also one that engages the public in a new way.
What does an average work day consist of, from when you arrive in the office to when you leave?
We arrange for our work day here on Earth to match the Martian night, so when the solar-powered Rovers shut down for the Martian night, they send us pictures and other data showing us where they are now and what they did the previous day. Then they go to sleep, and we go to work planning the next Martian day of activities. At the end of our day, we have a list of commands to send the Rovers when the sun comes back up in the Martian sky. At that point, they go to work, and we go to sleep.
The Martian day is about 40 minutes longer than an Earth day, so this affects our work schedules: some days we start earlier, some days later, and sometimes we plan two Martian days at a time for the Rover and skip every other day here on Earth. (We do that when the Martian night starts so late in the Earth day that there isn't enough time to plan a regular day of activities.)
Insofar as there's an average work day driving the Rovers, it looks like this: as fast as possible -- usually within the first half hour -- we assess the results of the previous day, make sure there are no concerns, and work with the science team to make a good guess as to what we should try to achieve that day. Over the next hour or so, we refine that plan, until we have a solid high-level idea of what we're trying to accomplish -- drive over to that nifty-looking rock about 20 metres away, or pick up the Rover's arm and put it down again over there for example.
The rest of the day involves turning that high-level plan into detailed commands that the Rover can execute. First, we spend an hour or two putting together a good enough version of those commands so that we can show the whole team an animation. Then we hold a meeting where all of the engineers and scientists approve the detailed plan and review the animation together.
In the last phase, we spend a few more hours refining our first cut, debating what-if scenarios. What if that patch of soil is deeper than we thought and we swerve off course while driving through it? Or what if we start to go off course and whack the solar panels into that big rock? We make sure that the commands we're sending to the Rover will do the right thing in the face of all imaginable contingencies. We review this final cut at the day's commands -- twice! -- and if we can't find anything wrong with it no matter how hard we try, we send it to the Rover.
The next day, it starts all over.
What has been the most exciting or memorable moment working as a Rover driver?
My first drive will always stay with me, of course. It wasn't much, just seven metres, but when I went home I simply couldn't sleep. I realised that right at that moment, a hundred million miles away, there was a robot on the surface of another planet doing what I had told it to do. I still have never gotten over that feeling.
But for me, the greatest moment of the mission was when Spirit reached the top of Husband Hill. We took our little Rover -- just the size of a golf cart -- and drove it all the way to the top of a hill that's about as tall as the Statue of Liberty in New York. Spirit had to climb that hill to look for evidence of water because most of the crater it landed in had been covered by a lava flow billions of years earlier, hiding any possible water evidence. Our hope in driving Spirit up Husband Hill was that it'd discover rocks that had been altered by water but not subsequently covered over by lava -- and it did just that, almost as soon as it started climbing.
But it didn't stop there. Plucky Rover that it is (and plucky team that we are), it kept on going, all the way to the top, even though it took months and it suffered some heart-breaking setbacks on the way -- times when it had to give up hard-won altitude so it could attempt a different ascent path. As hard as that was, it was worth it. Coming in one morning and seeing the images from our latest drive, realising that our courageous little robot had fought its way to the top at last, filled me with unending pride in the robot and in our team.
What has been one of the more humorous moments at work?
Early in the mission, we nearly lost Spirit due to a problem with its flash file system -- in PC terms, it couldn't boot normally because we'd filled up the Rover's hard drive. (Instead of a hard drive, the Rovers use flash drives -- essentially the same technology that's used by the USB drive on your key chain -- because they have no moving parts.) When we had diagnosed and fixed the problem, cleaned up the flash drive, and knew that the danger was safely past, someone wrote this on one of our white boards: "Spirit was willing, but the flash was weak."
What has been the most challenging aspect of your work, technically?
Working around hardware failures. The Rovers have now survived more than 15 times as long as they were built to, in an extremely harsh environment. There's lots of radiation, fine dust lurks everywhere, and temperatures swing from mild to Antarctic -- often in the same day. In the face of this, some of the Rovers' parts have just plain worn out, and there's obviously nobody around to repair them (though I'd go if they'd send me!), so we have to muddle through.
A perfect example is when Spirit's right front wheel seized up just before the onset of the previous Martian winter. Imagine trying to cross the desert pushing a shopping cart with one stuck wheel -- suddenly, that's what driving Spirit was like. We couldn't get to our planned safe haven for that winter -- it was on the wrong side of a sandy patch that our hobbled Rover could no longer cross.
So now we have winter coming on. The Rover's going to die if we can't get it to safety, and we have to learn to drive it in a whole new way. Thankfully, we have a copy of the Rovers in a testbed here on Earth, and we use it for just such occasions. We went down to our testbed, invented techniques to drive Spirit successfully despite the hardware failure -- with Martian winter edging closer all the while - then found a relatively nearby spot where we could park it safely, and somehow managed to drag it there.
That was a stressful time, but it was also enormously rewarding. The guys who worked out those driving techniques saved a priceless scientific asset. And even better, because Spirit survived that winter, it ended up making the biggest scientific discovery of the entire journey the next Martian spring, discovering a patch of silica-rich rocks. Silica-rich rocks thrill our science team because they're excellent evidence for past water exposure -- exactly what we're on Mars to find.
What Spirit found was probably a hot spring or fumarole, the kind of place where biologists think life might have arisen on Earth. It's not equipped to detect evidence of past life directly, but it found exactly the sort of spot we want to go back to with robots (or people!) that can.
Oh, and the first hints of that silica-rich material were turned up by the trenches we dug as we dragged our broken wheel around. Without that hardware failure, Spirit wouldn't have realised one of its greatest successes!
What was your technical background before joining NASA as a Rover driver?
I completed my Master of Computer Science degree at the University of Illinois at Urbana-Champaign, and I have undergraduate degrees in Computer Science as well as English. The degree in English surprises a lot of people, but it's not so unusual on this team -- other Rover drivers on the team have undergraduate degrees in fields as varied as anthropology, sculpture, and chemical engineering.
I came to Jet Propulsion Laboratory (JPL) straight from grad school in 1994 and worked here for about five years, working a little bit on each of a lot of missions, before I landed a full-time software development role on the mission that became Mars Exploration Rovers (MER). That role was to help write the software we'd use to drive the Rovers, and I parlayed that into the Rover-driving job.
What software and hardware do you use to do your job?
Since this is a one-of-a-kind endeavour, the software is almost all custom-built, as you can imagine. We wrote software for driving the Rovers, a collection of tools known as RSVP -- the Rover Sequencing and Visualization Program. We use a lot of other custom-written software, written in all kinds of computer languages -- Java, Perl, Python, shell scripts, you name it. When we need more, we write it.
It all runs on a collection of high-end Linux boxes -- nice systems, but commodity PC hardware. Since 3-D visualisation is a big part of the job, the Linux boxes sport bonzer NVIDIA graphics cards.
In order to plot the Rover's next moves you must need to "become a robot", in the way an actor needs to become his or her character. Do you find that you think more like a robot in your everyday life from doing this job?
That happened to me very quickly. Just a few weeks into the mission, I was walking across the JPL campus, and I realised I was half-consciously evaluating the rocks I saw as if we were planning to drive near them: "No problem, we can drive over that one; oops, that one's too big, better steer around it".
That evaluation is a large part of what we do in our job, and it was quite funny to realise that it had seeped into my real life.
Would you describe your job as a "dream tech-job"?
Yes. Oh, my, yes. I've often said that I have the best job on two planets, and you can believe it.
Read the interview with Ashley Stroupe here.