2009-02-28

Spirit Sol 56

[This originally happened on 2004-02-29, but since that year was a leap year and this one isn't, I'm going to make my life simpler by posting this on 2009-02-28 instead, and sticking with the original dates starting with the next sol (on March 1). So you get two sols in one day. Happy bonus sol!]

It's a mess. Last night, after I left, the mission manager got paranoid about the drive. Specifically, the concern had to do with the fact that the drive had been broken into two sequences, one to drive back and another to drive forward. What if the backward drive never got uplinked correctly? Then the rover would ram the rock.

Now, I admit I didn't think of this objection, but I wouldn't have solved it the way they did. They could have merged the two sequences into one, or had the first sequence invoke the second (hence the second would not run without the first). Chris Leger later comes up with a third way: send the rover a command that ground-precludes driving, then clear that flag from the first sequence.

They didn't do any of those things. Instead, they redid the forward drive in hazard-avoidance-enabled mode.

But their target is a big rock. Which looms as a hazard in the rover's on-board navigation map. So it will think we're telling it to drive into a rock with hazard-avoidance mode on. When you do that, it thinks you're crazy. And it stays put.

The upside of having lots of other people look over your work is that they'll find problems you won't. They've done that lots of times, enough times to make this a small price to pay.

But ... damn it!

We at least got the RAT mosaic. We can already see it in the images, a Mickey Mouse head with indistinct upper ears -- their upper halves fade out, showing where the uneven rock face curves away.

We got something else, too, three big anomaly reports. It so happens that there's no way to make your command sequence conditional on whether the RAT contact switches have tripped, indicating a successful placement. So normally, we have to place the RAT, get an image of its placement, review that image on the ground, and decide whether to proceed. However, each go/no-go cycle costs us a sol, and we wanted to do three RATtings in a single sol. So Bob and Eric came up with a tricky set of commands that get the desired effect. Your sequence looks like this: you place the RAT, make sure the IDD is happy, then issue a command that happens to generate a fault when the contact switches are tripped. Then you check to see whether the fault occurred. If it did, that means the arm is actually OK but the switches are tripped -- which is what you want, so you clear the fault and go on.

It's a clever idea, but nobody considered that the arm was going to think it faulted out three times. Every time it faults out, it generates an anomaly report, and these chewed up a good chunk of our downlink. Oops. If the drive had completed, that would have really hurt us. As it is, our other screw-up masks this one.

I'm feeling bad about the drive screw-up. I can't decide whether it's my fault, or one of those things that sucks but just isn't anyone's fault in particular. I've been splitting sequences up to make life easier for the SIEs, but if I hadn't sequenced these two drives separately, the issue wouldn't even have come up. But now's not the time to worry about it -- maybe we can recover. We talk about whether we can build and uplink a short drive sequence. It wouldn't be difficult, but if we do this, we'll need to do it before Earthset. When will that happen? Andy looks at his watch and thinks. "Right about ... now," he says.

Great.

When I get to the downlink assessment meeting, the SOWG chair for the day, Ron Greeley, is telling the science team that the science plans have been aggressive lately and the uplink crews have been great about supporting them. "When things don't go well, don't rag on those guys -- it ain't fair." This is true, and a nice thing to say. Perversely, it also makes me feel worse: I'm not just disappointing the scientists, I'm disappointing very nice and fair-minded scientists. Can't they just be assholes?

There are still possible recovery strategies. Maybe tomorrow we could plan a short drive for the morning, then come in in the middle of the day, get the telemetry, quickly sequence the RAT grinding, and let the grind run in the afternoon. This would save us a sol over the naive plan, which does the bump-drive and the grind on two separate sols. If we can't get the grinding sequence done, then we're no worse off. I quickly volunteer to try it, aware that this means I'm volunteering for something like two hours' sleep tomorrow. If that's what it takes, that's what it takes.

But will it work? Do we have enough time? My first impulse is to say yes, we can do it, but Andy wisely gives me a half-hour or so to think it over. The lead RAT, Steve Gorevan, and I get together to talk about it, and when we think we know what it will take, we talk it over with Andy and Mark Adler. ("Just say no," Adler half-jokes the minute he sees our worried expressions.) Once we get the post-drive downlink data, it will take maybe half an hour to get the terrain meshes and other products. Then the RAT guys need to find a suitable target on the rock. Then I need to take the mesh and the target and write a sequence that does what they want. We need to simulate it, test it, review it. There's no way to do all of this in less than a couple of hours. Then we need to actually uplink the sequence, and the rover needs to carry it out. The RAT grinding itself takes four hours to do properly on a rock that's as hard as this one. So we're looking at needing to have six usable hours left in the sol, and that's optimistic. We probably won't have that much time, so they'll need to shorten the RAT grind, which is an idea nobody likes.

But it might work. So the four of us troop into the downlink assessment meeting, which is still going on. ("This is troubling ...." says one of the MTES PULs when she sees us.) "You have an answer!" says Greeley. "We have questions," I respond. We lay out the plan, with all of its caveats. They don't seem to like it. They want to MI the pre-grind area, and if I'm honest with myself, I have to say this will add complexity that's likely to push us over the limit of what we can do in the time available. I still want to say we can do it, but I do the smart thing and stop pushing. Greeley gently nixes the idea, and Arvidson concurs: "Do it right!"

So this will be a slow day. Or not. Steve and another RAT guy, Tom, point out that we'd have less trouble hitting targets on this rock if we maneuvered around a little so we were facing it dead on. From where we are (or, to the point, from where we will be when we complete the drive with a short bump tomorrow), the reachable face of the rock slopes away to our left, an awkward surface for us to try to reach. So now I have something to do: see whether we can get the rover a wee bit to the left, so the rock will be easier to poke at.

The more I investigate the idea, though, the less I like it. This rover can't just crab-walk; if you want to be half a meter to your left, you need to do multiple turns and back-and-forth drives. It's not that complex, but given how the last couple of sols have gone, I decide I'd prefer to stick with the devil we know. Targeting the rock won't be easy, but we know we can do it from where we'll end up.

While I'm looking into this, Sharon Laubach pops in to say hi. I was right, she is changing jobs. But I was wrong about which job. She's going to be a rover driver, probably for the extended mission. This is a good thing for us and makes her very happy, but I can't help asking whose shifts she expects to pick up. She's vague about it. Maybe she'll spell me for some of my shifts if I have to write some software to support extended ops, she suggests. I comically hunch over and hug the monitors possessively, and she laughs. But I wonder if this means I'm not going to be able to drive the rovers in the extended mission, or if my opportunities to do this will be reduced. I guess we'll see.

It's looking like there's no longer any hope of reaching Bonneville by sol 60, which was our plan before the scientists got distracted by the shiny new rock. But in all fairness, that's the nature of the tactical process: we're supposed to leave our plans flexible, so we can react to the changing landscape.

Andy manages to put a good face on our situation at the SOWG, pointing out that tomorrow we'll manage to get a full charge on our batteries -- we normally do this once per week anyway, and it's been 10 sols since we did it last. Speaking of recharging batteries, the RAT PUL looks exhausted -- they've all been working extra because both rovers are RATting at about the same time. Steve Gorevan asks if I need him for anything, and looks very relieved when I tell him no. He leaves to get some sleep.

The plan for tomorrow is very simple. We're used to constructing plans with lots of things happening at once -- doing remote sensing while the APXS integrates, for instance -- but tomorrow is pretty well serialized. Apparently, this is how our extended-mission plans will tend to look.

I don't have much to do, since tomorrow we're just doing a simple 55cm drive straight ahead. I finish the sequence in minutes and drift into the science talks, which are now being held after the SOWG instead of after the downlink assessment meeting (to give the scientists more time to prepare for the SOWG, I suppose). Today they're discussing the trench results. Jeff Johnson shows the MI trench images, comparing the top and wall of the trench with the floor. The APXS guy reports that the spectra of the floor and the wall are very similar to each other, which could indicate water-driven soil mixing.

The free ice cream is now gone. Andy seems to think this is a bad idea, pointing out that the ice cream was meant as a recognition that we were doing something a little extra -- by which he means living on Mars time. When the topic has come up in meetings, he habitually declares that "if the ice cream goes, we no longer have to work Mars time." I don't care about the ice cream and I love working on Mars time, so I don't say anything, but I'm sympathetic: working on Mars time does disrupt your life. Mark Adler seems to be on Andy's side: he's filed an ISA (roughly a system bug report) on the lack of ice cream in the freezer.

I'd go home early but I decide to stick around for the midnightly science talk. This gets delayed and delayed, but when it finally happens it's way cool. John Grant tells us why Opportunity is finding El Capitan so interesting: the short version is, multiple converging lines of evidence all point to the rock being formed in the presence of water, probably a standing body of water that covered a large area. This has broad implications: Mars as a whole must have been warmer, and must have had significantly greater air pressure, at some point in its past. The larger implication of that is that conditions were likely favorable for the existence of life.

They'll announce this to the world on Tuesday.




Courtesy NASA/JPL-Caltech. We've now drawn a Mickey-Mouse head on Humphrey, by brushing away light-colored dust in three spots to reveal the darker rock face beneath. Essentially, it's graffiti on Mars. I get paid for acts of interplanetary vandalism.

1 comment:

Anonymous said...

"I get paid for acts of interplanetary vandalism" has entered my quote file.