2009-02-23

Spirit Sol 50

The MI images we took of the trench wall don't seem to be in focus, blast it. One entire series is now down and is definitely out of focus; only the last image, the one farthest from the trench wall, is nearly focused. The other series are incomplete, but we don't have any in-focus images yet, and we don't want to leave without getting some nice closeups of the trench wall. So we're going to take a last shot at it before we leave, retaking a few images of the same spots from a little farther out. We're not sure why the images are unfocused, but the IDD doesn't seem to be positioning itself in the exact desired position relative to the trench -- whether because of a problem in the IDD itself, or because the terrain meshes are inaccurate, or for some other reason. Still, we deliberately touched the wall yesterday when placing the MB, and the IDD can remember that position and use the known position for targeting the MI, at least on that spot. This gives us some hope that the new pictures will look better.

After we do that, we'll drive backward about a meter, MTES the trench, and move on. The only complication is that we're also being asked to take MI images of the capture and filter magnets. These magnets are located on the rover deck, at the base of the mast. Getting the IDD to place itself there is tricky, and not something I've done before. We have a canned sequence to do something similar, but not identical, and given the potential risks, I decide to warn them we'll need to do it in the testbed first. But Ray Arvidson puts the kibosh on the idea anyway, at least for tomorrow. It turns out that this qualifies as a first-time activity, and we have rules about those -- one of the rules being that you can't do a first-time activity and drive on the same sol. At least, not without ground confirmation that the first-time activity completed successfully, and this will cut significantly into our drive time. And Ray wants to drive.

So Ray plays bad cop. The guy who wants the experiment is pissed off, but Ray just tells him, quietly and flatly, that it's not going to happen tomorrow. It'll happen in a couple of days, when we get to our next target. This gives us time to test it in the testbed and makes sure we keep moving for now. The requester is mollified slightly, but still leaves in a huff. Ray is Ray, unperturbed.

While I'm working out the sequence we'll need to test, Bill Dias stops by to talk to me. He's grumpy about something, and I decide to lighten him up a little by teasing him, telling him that they should just put him in charge of the scientists. "I'm glad you feel that way," he drawls. "They should put me in charge of a lot of things." You have to know Bill to be sure, but it worked; he's relaxed.

At the downlink assessment meeting, LTP lays out the plan. Tomorrow will be a touch-and-go sol, with a short drive of about 10m. The following sol, we'll shoot for another 40-50m, to get us to our next target, a spot called Middleground. Middleground lies smack between the two locations the scientists are really interested in, Humphreys and Sandia; when we get there, they'll decide which of the two targets -- or both -- they want to examine more closely before continuing to Bonneville.

LTP also has a warning: we're putting more data into flash than we're taking out, and this trend obviously can't continue indefinitely. In the near future, we may have to cut science observations in order to give the rover a chance to clear out the backlog. We have to think of it like this: lower-priority science observations we take today might trump higher-priority science observations we'd like to do a couple of sols from now.

One science observation we did get yestersol was a series of images taken over the span of a couple of hours, showing the rover's shadow lengthening over the Martian ground. This was accompanied by MTES observations showing how the rover's shadow affects the surface temperatures, but the movie itself is a cool outreach product. Ray had been against this as a waste of time, but now admits to the experimenter, "It's good that you fought for that."

Another result is reported, this one from the APXS. The trench floor looks just like the surface to the APXS -- in other words, digging in the soil doesn't reveal a significantly changed composition. They're guessing the wall measurements will show the same results, but the results aren't in yet.

Our drives have a new wrinkle: an antepenultimate step. We've long had a pair of steps at the end, a penultimate and ultimate, which let us take end-of-drive images that ensure that it's safe to deploy the IDD. The rover can't see the area immediately under the IDD, so if we had only the ultimate images, we wouldn't be sure it was safe to unstow the arm. Instead, we stop at the penultimate position, take an image, then take another step forward to take the ultimate image; this leaves us with images showing not only where we ended up but also what's underneath us. And now we have a step before that -- the antepenultimate -- to MTES and PANCAM the IDD work volume, which can't be done from the penultimate position (it's not back far enough). Ray hates this, both because it complicates things and because he can't pronounce the word. He keeps pronouncing it "antIpenultimate" -- as if it meant "against the penultimate" -- instead of "antEpenultimate," meaning "before the penultimate." Every time he does this I want to correct him, but I manage to restrain myself.

This meeting has long been run with just a pair of hand mikes, which as I've said before influences how well it works. An interesting change today: Ray wears a throat mike, so he can talk whenever he wants, and they hand the other mikes around as before.

Nathalie Cabrol gives the scientists a similar talk to the one she gave in the SMSA yestersol, regarding the polygonal soil structures and what they might mean, and their connection to the trenching experiment. The polygonal structures hint at freeze-and-thaw cycles, she says, and looking in the trench may help to confirm or disconfirm this hypothesis. Freeze-and-thaw cycles tend (on Earth, at least) to result in a mixing of soil layers, so that's one thing we'll be seeking evidence of as we inspect the results from exploring our trench.

At the end of the meeting, one of the MTES guys shows a HyperDrive-produced shadow movie, illustrating what the shadowing will be like tomorrow, to help them plan out the timing for tomorrow's science observations. Ray is impressed by this and emphasizes the importance of using our modeling tools for these purposes.

The SOWG goes relatively smoothly again; once more, I get a decent amount of the sequencing done before the meeting ends. But I do discover some bad news: we don't have as much of the surrounding terrain model available as I thought. We have a wedge directly behind us, and good coverage in front (the planned drive direction), but nothing to either side. Normally, this would be fine, but today we have something new in front of us: a trench. I had planned to drive around the trench, but without knowing what's on our right and left, I can't safely do this. What's more, we can see just enough of what's on our left to know that it's almost certainly not safe to drive there. We could zoom around the trench to the right, but it's a four-hundred-million-dollar gamble, and I don't like the odds.

So I look at another option: just drive straight ahead, right over the trench. Why not? The trench is only about 6 or 7cm deep anyway, and not so broad that the wheels are likely to get stuck inside it. I model this in HyperDrive for a while, and slowly start to think it might work. HyperDrive lets you see the rover's own evaluation of the terrain, and the rover doesn't see the trench as an obstacle. Then again, the rover's terrain evaluation is intended mainly for rocks, and for those it performs superbly. It doesn't always do as well with slopes -- or holes.

And it's another one of those things I haven't done. I deliberate for a while, and convince myself that it's liable to work out just fine. To be on the safe side, I end up calling Jeff Biesiadecki on his newly acquired cell phone (waking him, poor fellow). He hasn't tried it either, but thinks it should be fine, and there's no more expert opinion than his. So I decide to go for it.

The IDD sequencing complete, I put together a roughly S-shaped candidate drive. This first backs away from the trench so we can MTES and PANCAM it, then goes straight ahead, over the trench, zigzags between some rocks to the north, then bears northeast toward Middleground. At the end of the drive, after our terrain mesh runs out, I place a waypoint at Middleground itself, so that the rover will drive there on autonav if it gets that far. The amount of drive time we have available keeps changing as they refine the plan, and I have no idea whether we'll get more than a couple of meters of driving, much less the 18m or so that I've planned out -- and we almost certainly won't even get started on that final waypoint -- but I figure I might as well plan the most aggressive drive I can. If the rover runs out of time, it will just stop wherever it ends up, and if it gets enough time, well, we might as well cover as much ground as possible. The stated goal of the drive is to take us 10m closer to Middleground, but the real goal -- I am told in semi-joking confidence -- is just to get 1cm away from the trench, so the scientists will stop poking at it, and drive on already. This, I'm fairly sure we can do.

Sharon Laubach is hanging around, looking bored. Opportunity has a very tight IDD maneuver planned today, one that's making everyone nervous. They're placing the APXS up against El Capitan, and the positioning just happens to put the side of the RAT right up against our forearm, with maybe 2mm of clearance. If the positioning uncertainty is too great, the maneuver will cause the turret to rotate just a tad too far, straining against the forearm itself, and this will likely fault out the sequence. (Frank asked me to look at it earlier, and I didn't like it any better than he did. The software says it's OK, but we're not so sure.) They've lost a couple of sols lately and really don't want to lose another, so Sharon has come downstairs to fetch Bob Bonitz, the foremost IDD expert, to take a look. But his shift hasn't started yet, so she's just zoning until he arrives.

"Hey, you want to help drive the rover?" I ask her. Sharon's the honorary ninth rover driver anyway -- might as well make her work for it. She accepts the invitation, and we look over the drive together until Bob arrives. The drive itself isn't all that complex, but I don't know how the rover will slip when driving over the trench (it's not just the hole itself I'm worried about, but also the loose soil that we dug out of it). So I keep refining and refining that part, adjusting the drive distances until the rover is clear to turn without falling back into the trench, placing a waypoint to force the rover to automatically correct for slip. I realize I've started to understand the rover's behavior at deeper levels than I used to, and this is helping me build up a decent bag of tricks. I also add a command to dump the navigation software's memory usage statistics, as Mark Maimone has requested. It won't do any good unless we get to the autonav, but it won't hurt.

The other person Sharon has been looking for shows up, so I promise her I'll send Bob up immediately when he arrives. She goes off to talk to Richard Cook. (Richard was recently made project manager when Pete moved on to MSL.)

When Bob shows up, I send him upstairs as promised. He doesn't return until hours later, with the news that he told them to go with an alternate sequence. He'd done a similar move in ATLO, and found that the IDD had even less clearance than the software is now showing. This, together with the positioning uncertainties we have to deal with anyway, convinced him the risk was too great.

He's also a little annoyed that they didn't just call him at home. "They had a few hours of margin, and I think they didn't want to wake you," I tell him. He shrugs and says, "Better they should wake me up than lose a sol. What's a little sleep?"

Bob tweaks the IDD sequence I'd written earlier (maneuvering in the trench is tricky, and he has some ideas on improvements) while I finish up the drive. At the walkthrough meeting, we come up with an idea: we should take a picture looking back at the trench after we drive over it. This has very little science value but will be a cool outreach product, so the idea wins. It's not fully clear whether we have management's approval, but we think it's a cool idea, and Ray thinks it's a cool idea, and the guy who would have to build the imaging sequence thinks it's a cool idea, so we just go ahead and do it. (As a safeguard, we make the image a low-priority data product, so that it won't push out any science.) I'm still not sure how far we'll slip, so I don't want to put this in the drive-over move itself -- if the rover gets stuck in the trench, the situation isn't going to get any better while it sits there for a full minute taking a picture. I want it to keep moving. So I add the command to take the picture from a little farther away; the picture likely won't be as cool, but we'll get it.

This has been a long day for me, I realize; because they pulled Bob away for several hours, I figured I should pick up some of his shift, and the result is that I've been here something like 14 hours. But working together, Bob and I get everything done, and only about 5 minutes late. Not too bad, I guess. I finish up the (electronic) paperwork for the drive sequence and go home.




Courtesy NASA/JPL-Caltech. Maneuvering the IDD in the trench -- tricky business.

2 comments:

Jess Robinson said...

ATLO isn't in the glossary, what is it?

Great blog!

Scott Maxwell said...

@Jess Sorry about that! ATLO is Assembly, Test, and Launch Operations. I'll add it to the glossary. Good catch, thanks!