2010-09-30

Captain's Log, Supplemental

And so it was just about this time, five years ago, that Spirit took the Husband Hill panorama -- a beautiful color panorama taken from the very summit of Husband Hill. (It's also called the "Everest panorama," and sharp-eyed readers might notice it's the image at the very top of this blog page.)

The rovers have returned more than a quarter of a million images, but this one would be on my list of the top five or so. There's so much in this image. First, and maybe most importantly, are Spirit's own tracks leading to the summit, a record of the struggle and hardship we faced on the way up -- and the relentless determination we showed in its face. In the distance, there's a dust devil on the plains, caught in the act of moving between frames. (So it looks like two dust devils, but it's really one, moving as Spirit took her images using a series of three different-colored filters.)

And, of course, there's Home Plate, to whose immediate west Spirit sits today, trying -- I hope, I fervently hope -- to phone home.


The Husband Hill panorama. (The link takes you to the full -- very large -- image.)

2010-09-29

Spirit Sol 619

Coincidentally, my last drive was performed without the assistance of the ultimate FHAZ, and so must this one be. Spirit's just a few meters away from the true summit of Husband Hill -- a slightly higher point than the spot we last thought was the summit -- and we're going to drive up onto it.

That's a given. The big discussion at the SOWG meeting is what we're going to do next -- I mean, over the weekend. Should we pick a juicy rock and IDD it, and so use that rock as thisol's drive target? Or should we just get into a good position to do imaging, and do self-IDDing (such as by putting the MB on the CCT) over the weekend if we IDD at all?

Hap's take: "We're too experienced to be wasting our time exploring dusty rocks. We should wait and do clean rocks somewhere else."

Squyres agrees: "We may have to trade this strike-and-dip imaging against the IDD position." But they can't get the pictures of rocks they want anywhere else, while the rocks they have pictures of are not particularly interesting IDD targets, so that's a bad trade. "So let's do structural imaging on some face of Hillary" -- named for Sir Edmund Hillary, of course; another rock up here is named for his Sherpa guide, Tenzing Norgay -- "and a panorama."

Alian suggests that we could get into position for imaging and IDD whatever's there, but Squyres says no. "I'm reluctant to use the IDD for low-value targets. Maybe we could get into a good position for imaging and recapture the MB-on-the-CCT measurement we recently lost." And that's pretty much it.

John handles most of the drive, with me looking over his shoulder and offering feedback as he goes. This is a practice we sometimes call "backseat rover driving," and it's a really effective one. I never feel like I'm contributing as much when we plan drives that way, but I'm actually being more useful than if I make the RP-1 do everything himself and then get involved only later.

As we review the drive at the walkthrough, Steve chimes in to remind us of what a big deal this is. "This is the final drive to the summit of a mountain we've been climbing for more than a year," he observes. "And it's our last real uphill drive for a while."

This should feel like a more momentous occasion than it does. I think we're all just kind of fatigued. Morale isn't what it once was. And that's a shame, because we put a lot of hard work into getting here, and we ought to feel damn proud of that. I'm sure we do, really. It's in there somewhere, behind all the exhaustion.

[Next post: sol 625, October 6.]

2010-09-23

Opportunity Sol 592 (Spirit Sol 613)

I've been out for a while because of my shoulder surgery, and now, for the next six weeks, I'll have to drive the rover with my arm in a sling. My right arm is barely usable at all -- can't really type with it. Fortunately for me, we've got a relatively easy sol. Opportunity's parked on some outcrop called "South Shetland," and a couple of meters away is a chunk of dark rock called "Deception." We've just got to bump to it, to set ourselves up for a weekend of IDDing.

And, of course, it's not that simple. For some reason, the downlink was unexpectedly meager, and we missed most of the ultimate FHAZ. Fortunately, Cooper's able to work it out using previous sols' imagery and the penultimate imagery from this drive, so we'll make it. With a small dash of luck, anyway.

I'd cross my fingers for luck, but my right hand's not up to it.

[Next post: sol 619, September 29.]

2010-09-17

Spirit Sol 607

The scientists are happier than I am. In and of itself, that's an okay state of affairs; I'd rather have it that way than the reverse.

The scientists are happy because they have a nice scuff to play with. I'm unhappy because the scuff isn't where it was supposed to be. It's on the wrong side of the dune -- the far side, not the near side. This is an error of maybe half a meter; we should have been a lot more precise than that.

Chris works out what happened. In the main drive sequence (the "backbone," as we call it), we had several clauses that were intended to help us execute the drive more precisely. They checked if we were close to our defined target, and if not, they commanded a short step in its direction.

The problem arose because of our obstacle-check helper. For that part of the drive, we were running the helper explicitly as needed (as opposed to just letting it execute in the background) because the same "variable" that it uses to check whether we're too close to each obstacle is also used to check whether we're close enough to a goal. We wanted to be sure that the helper and the backbone didn't stomp on each other's use of this shared resource. The problem that escaped both Chris's and my notice was this: the backbone didn't reset the variable's value after calling the helper. So the helper stomped on the value, then we neglected to reset it to what we wanted.

The outcome we got -- a perfectly scientifically useful scuff half a meter away from the intended position -- was just about the least bad thing that could have happened. This mistake will be the subject of an automated check very shortly.

But not as soon as I'd like, I'm afraid. I'm going in for shoulder surgery Monday and will be unable to work for at least a couple of days. Chris is already covering my Monday shift, and John agrees to take my Tuesday shift. I already had Wednesday off, so I plan to just take those days off altogether and come in Thursday to help Jeng drive Opportunity -- one-handed, for my part.

It looks like I'll be leaving the rover in some increasingly capable hands. Since I had to meet with the surgeon and have some labwork done this morning, Paolo filled in for me until I was able to get here. I was a little worried about it since we're planning IDD work thisol, and Paolo has told me he's a lot less comfortable with the IDD than he is with driving. But I ask him to walk me through the sequence, and he knows it just about perfectly. He's even able to sketch out the process of doing it; he'd probably have been able to build the sequence himself, or come close, if John hadn't been here to do it.

Impressive.

[Next post: sol 613, September 23.]





Courtesy NASA/JPL-Caltech. The scuff, right in front of us.

2010-09-15

Spirit Sol 605

The Erebus Highway drive worked splendidly, putting us right where it was supposed to. Now, back on the other side of the world, I get to help pull off another miracle: an 11m drive followed by a scuff. This is just like we did on Serpent way back on sol 72 or thenabouts, only starting from much farther away. This should be a good test of our abilities: have we learned enough since then to make something like this work? Chris is more confident than I am, but I sure hope he's right.

We break down the responsibilities more or less like this. Chris is going to worry about getting us there, and Ashley and I will work out the obstacle-checker helper sequence and the scuff sequence itself. The obstacle checker is going to have an important responsibility thisol: keeping us from driving over a cliff. Really. The dune (ripple?) we want to scuff is right on the edge of a cliff, and if we go a meter or so past it, it's bye-bye, Spirit.

The way the obstacle checkers work is this. We work out a list of known obstacles, and for each one, we define a circular hazard region around it. Essentially, the obstacle is the center of a circle we want the rover to stay away from. Then we plan a path that avoids all the circles, but just in case, the obstacle checker runs in parallel with the drive, periodically checking off its list and making sure the rover is outside of each circle. If the rover strays inside a circle, the checker issues a command that stops the rover from driving any farther.

While there are several other obstacles, the most interesting, and the one I attack first, is the cliff itself. When I rotate the terrain mesh so that we're looking down on it, I notice that the cliff face is a rough semicircle. Hmmm. Working by eye, I estimate the center of that circle, pick a radius -- eight meters -- and ask RSVP to show the circle I've just defined. It matches up with the cliff face almost perfectly on the first try.

"Not gonna get much better than that," I say nonchalantly and go on.

For the scuff, we have a couple of choices. We've done two basic kinds of these. The first was the one we did on Serpent, where we started by planting a wheel where we wanted the scuff to begin. We then turned the rover about five degrees to the left, then ten to the right, then five to the left, leaving it where it started. Next, we backed up 10cm and repeated the wiggle maneuver. The result was a beautiful, broad and deep scoop in the ripple face.

The other kind of scuff we've done didn't involve a ripple; we were just scraping off a top layer of soil. That worked differently. We spun both front wheels forward, holding the other wheels steady, to scuff away any material directly under them. Then we drove the rover a few cm backward with the front wheels locked in place -- using the middle and rear wheels to drag the front two. Then we repeated the whole maneuver, spinning the front wheels to kick away any material that accumulated under them while we dragged them and then dragging them back again, and so on. The result was just what we wanted: a shallow, clean scuff that exposed the immediate subsurface layer.

In this case, we're after the deep scoop, so I pick the first one. This is quite a blast from the past: I vividly recall building this scuff sequence the first time, and though we have to modify it slightly for this new situation, I have a lot of fun revisiting it.

So the sequence is built, and we cross our fingers that this very aggressive go-and-scuff will work. As it happens, today comes with a reminder that Murphy's Law is very much in force. They missed yesterday's downlink on both vehicles due to some kind of DSN station problem, and today we find out the details. It turns out that rodents chewed through some cables at one of the antennae -- the one our downlink was coming in on -- at the DSN complex in Goldstone.

Chris has the perfect pun for the occasion. "Rats!" he exclaims.

[Next post: sol 607, September 17.]

2010-09-10

Opportunity Sol 582 (Spirit Sol 600)

Today is Erebus Highway Day. I have a feeling about this. The last few drives have all been going great; we're really on a roll. And we can see the highway itself in the distance.

Wait -- "the distance"? Heck, that thing's only 25m away from here.

The problem is, we have only two hours to drive, and with our visodom-heavy drive structures, that's not a lot of time. But I look carefully at the images, and there's a straight path along mostly low ripples -- the highest point I can find is about 13cm, and that's nothing -- leading all the way to the highway. It's the on-ramp.

Well, Mars doesn't give us these gifts every sol, so when we get 'em, we might as well take 'em. I plant a waypoint on the highway and plan a drive that takes us straight there, with periodic slip checks. This is nothing -- should take only an hour and a half or so, even with some extra imaging we're taking along the way.

If all goes as planned, we'll be on the Erebus Highway by Monday!

[Next post: sol 605, September 15.]

2010-09-09

Opportunity Sol 580 (Spirit Sol 599)

Thisol's drive shapes up to be rather dull. We didn't get drive-direction PCAMs for some reason, so we're limited to the NCAM mesh, and this limits our drive to 15m or less.

But, of course, it's never quite that simple. Thisol there are two complications. First is the comm heading. We end this drive by heading south along a trench to a point where we hope we'll be able to cross a ripple on the next sol. Because of this southerly trajectory, the most natural heading after this drive would be about 200deg. But this is a terrible direction for comm. Good would be about 310deg.

On Spirit -- no problem. Turn in place. But on Opportunity, turns-in-place are limited. Paolo checks out an alternative drive, where we do a sort of K-turn before heading backward down the trench, but this doesn't cut down the turn's magnitude enough to be worth the extra effort and uncertainty -- not to mention that it converts the turn to a counterclockwise one, which is even less desirable.

After wrestling with this problem for a while, I realize we've overlooked something. The turning rules are relaxed when we're on outcrop -- and at the end of this drive, we'll indeed be on outcrop. So we should be able to do the large comm turn after all.

If we're sure we're going to be on outcrop, that is. We can solve that problem by using visodom -- but we don't have time to use it the whole way. In the end we use visodom for the first half of the drive, then do a blind segment with a visodom-based slip check, and then go the final segment blind. There's a little bit of finger-crossing involved in this, but not too much. If the slip check passes, we're making real progress, almost certainly enough to make it where we think we're going. And if we get it wrong, well, the turn in place is clockwise, so at worst, we should just be piling up dirt on the outside of the turn -- not scooping it into the wheel, which is what happens if we turn the other direction. And if we're really bogged down, one of the timeouts on the turn segments should trigger and keep the problem from getting too bad.

I think I'll cross my fingers anyway.

2010-09-07

Opportunity Sol 577 (Spirit Sol 597)

No rover driving today. Nobody thought to tell me or Cooper -- though, really, we could have figured it out if we'd looked at a (Mars) calendar. We won't even get the data for the drive they planned last Friday until 22:00ish today.

This is the latest (and I hope last) vestige of our anomaly-inspired paranoia. They wanted to make sure that all the other stuff they'd planned, which included some MTES activities, went as expected before we drove. And they wanted to avoid calling the team in over the Labor Day weekend. Plus, this week's only a four-day post-holiday week, and since we're in restricted sols, we can drive only twice this week anyhow. So they held the drive until the last sol of the weekend plan, and we'll hit the road tomorrow.

[Next post: sol 599 (Opportunity sol 580), September 9.]

2010-09-02

Spirit Sol 592

Just in front of Spirit are a couple of tasty-looking rocks -- concave, with some kind of striations, almost as if they were wheel prints. At the SOWG there's a lengthy discussion about them. At issue is whether we should stay and IDD them -- MI at least, possibly with APXS -- or drive on and possibly return for this later.

Driving away and coming back later is not something we do often. But apparently there's already one target they've pushed onto the return stack, some rock called "Irvine," and this could become another.

The reason they're eager to leave is simply one of timing. We've already performed the first half of a long-baseline-stereo observation, taking a bunch of pictures of the valley beyond Husband Hill. The other half of that experiment is where we drive about 10m away and take pictures of the same terrain from the other vantage point. It's like having two eyes set ten meters apart, which provides fantastic stereo coverage. And all this will be important for our future exploration.

But if we don't go ahead and do the drive, we'll be sitting in this spot for the weekend -- the extended, Labor Day weekend. The obvious compromise, doing a touch-and-go, is ruled out because of time and flash volume constraints. And while these rocks are intriguing, they're not worth spending five sols on.

That's particularly true in light of the continuing southward motion of the sun. In the next hundred sols or so, the energy curve will start to strongly favor south-facing slopes, which means we want to be climbing down the far side -- the south side -- of this hill before too much longer. And we have a lot of science and reconnaissance to do between now and then.

This all adds up to what should be an obvious decision: do the more important stereo imaging now, and come back to these nifty rocks later if there's time. But Ed Guinness, the SOWG chair thisol, is trying to get this across to the particularly wishy-washy scientists who are advocating staying, and he doesn't just take charge and push forward, as some other chairs would do in his place. So this discussion takes maybe twenty minutes, half an hour.

Eventually there's a compromise. All we have now is HAZCAM imagery, which isn't good enough to help us decide whether to return. So they'll PCAM the rocks before leaving, giving us much better data for that decision. Even this, however, turns into an extended discussion. How many filters? We don't have time for that many. We can't see it from the PCAM's right eye, so we'll have to back off before we use those filters and take the image mid-drive. Oh, now you don't want the right-eye filters? And so on.

When at last the time comes to name this target, one wag suggests "Indecision." But they don't call it that, of course.

What with the extended-mix vacillation, we almost miss the press conference. This is Steve Squyres and Chris Leger and Ray Arvidson and Jake Matijevic arranged around some guy I don't know, who turns out to be the NASA guy in charge of Mars exploration.

John turns to me. "Is Chris the only rover planner who's been in a press conference?"

"Yeah," I say. "I think it's because he's the only one who owns a tie."

Steve is up first. "Today is day 591 of our 90-day mission," he begins. He goes on to sum up the progress of both vehicles, emphasizing Spirit's arrival at the summit of Husband Hill. He shows off the first three-fourths of the 360-degree PCAM panorama from the top. And he and Ray announce that as Spirit ascended Husband Hill, she found definite proof of water in Gusev Crater. He reveals that we've returned ten gigabytes of data per rover so far. But the thing I think I'll always remember from this press conference is a picture Steve Photoshopped to show the scale of Husband Hill. He took the original color PCAM image of the Columbia Hills, cast around for some familiar object that was about the same height as Husband Hill, and put that object in the image in front of the hill. That object is the Statue of Liberty.

One of the last questions they take is for the NASA HQ guy: will the rovers continue to receive funding? Well, what can the guy say? "Yes, now that you've seen this spectacular image and we've learned there's irrefutable evidence of water, we're pulling the plug"? Of course not. Instead, he notes that we already have funding allocated through September of next year, with six-month reviews. "And as long as they keep doing good science, we'll keep funding them," is the substance of it.

John Callas, watching this with us in the SOWG room, pipes up, "Everybody write that down!"

[Next post: sol 597 (Opportunity sol 577), September 7.]





Courtesy NASA/JPL-Caltech. The front HAZCAM view showing the rocks we were thinking about exploring. I don't think we ever had time for them, though.





Courtesy NASA/JPL-Caltech. That hill was this high.

2010-09-01

Opportunity Sol 573 (Spirit Sol 591)

We're not finished being paranoid yet, so thisol's drive is still circumscribed. Callas had a 5m drive in mind, but a 5m drive would just take the rover a short way along a patch of outcrop -- it's hardly worth doing. We might as well cross the next ripple, and that's more like 10m. Somehow I manage to talk him into this, and over the ripple and through the woods, to Grandmother's house we go.

Er, or something like that.

I thought thisol would be a doddle, but it turns out to be a little more challenging than I'd expected. Not greatly so, but at least there's some original thinking required thisol. We're still copying from the sol 563/567 drive, and the next segment of that drive -- the one that takes us over the next ripple -- would leave us at a heading of 10 degrees. But that's relatively poor for comm; we need a heading of more like 65 degrees.

This change is not as simple as you might think. Back in the day, we'd just turn in place at the end of the drive. But on Opportunity, with her stuck right front steering actuator, the rules limit the size of point turns, this prohibiting this simple solution. Instead, we work out a way to modify Opportunity's heading before she crosses the ripple, which enables us to get her heading to 45 degrees. After she hops the ripple, she turns in place the remainder of the way.

This maneuver comes with problems of its own, though. Instead of leaving Opportunity heading more or less along the ripple, she's parked transverse to it. This is an uncomfortable position for a rover that can't do large point turns and can do sharp arcs only in one direction. But we keep plugging away, and manage to convince ourselves that a combination of sharp arcs and small point turns can get her back on track again. So we leave her there, with copious notes in this sequence about how the next one had probably better look.

The other thing that happens thisol -- well, first I have to tell you about my superstition. The two-weeks-to-Erebus thing doesn't count -- that's not a superstition, that's a law of nature. No, my superstition is that if you write the uplink report before the CAM, something will happen at the CAM that requires us to rebundle. Basically, it jinxes the CAM.

I've told Paolo (RP-2 thisol, shadowing Jeng) about this superstition, but he goes ahead and does it anyway. He has a good excuse -- he has to go help Mark Maimone get the SSTB rover back online -- but it availeth him not. Sure enough, during the CAM I notice an error in the RP sequence, which would cause us to record less drive data than we'd want -- part of our paranoia is that we want lots of this data, particularly when hopping ripples -- and we have to fix it and rebundle.

This superstition is well on its way to becoming a law of nature, too.