The good news is, Ashitey was right: we did make contact on the first retry with the tweaked surface normal -- none of the fancy new stuff was even needed. The bad news is, the resulting RAT brush was fairly poor, so they had to redo it last week. The good news is, the second one worked, and they drove away already.
And a nice job they did of it, too: 71m on the first sol, 47m on the second (autonav-only) sol. (Incidentally, this was Spirit's first such back-to-back drive.) The bad news is, the drive was sort of perpendicular to Home Plate, so our radial distance to our long-term goal remains more or less unchanged, even as time wasteth. Not that I'm worrying about it.
Anyway, as a result, we're only about 40m from El Dorado, and John and I are looking to get the rest of the drive today. Happily, it's just about as simple as it could be -- there are practically no obstacles here, just a nice flat zone. Just before El Dorado begins, there's a slight rise, so we'll have to be on autonav for the last part, but it shouldn't be any problem at all.
The big question is whether we should enter El Dorado or not. Our goal here is to find out what this thing is made of, which we intend to do by digging a trench. If we enter it thisol, then tomorrow we'll know a lot more about what it's made of and what we can expect from the trench. The problem is, maybe we'll know too much -- maybe we'll be bogged down in it, just as Opportunity was at Purgatory. And with winter coming on, and us already behind the curve ....
On the other hand, if we're going to trench the thing, we've got to enter it sometime, at least with the front wheels. Ideally, we'd stop with just the front wheels in it, but it's 40m away (or so -- our range data is poor, compounding the problem) and if we want to plant the front wheels only, we'd have an error budget of only 50cm or thereabouts. We just can't drive these vehicles that precisely. (They were designed for two-meter approaches. We do a lot better than that, but 40m really is out of the question.)
Fortunately, of course, we've got world-class experts to help us answer these questions. I ask both Brenda Franklin and Rob Sullivan about what we can expect from El Dorado. They agree that the material is coarse-grained (you can tell because it's dark -- dust is light). Brenda thinks the stuff is likely relatively compact, though maybe not; Rob's not even that sure. At least it's not likely to be quicksand -- so-called "foo-foo dust," which we'd just sink right into.
John and I decide we're going to proceed with caution. We'll just zoom right over the stuff we can see, but when we're about 10m from El Dorado, we turn on autonav and start doing slip checks every 2-3m (as opposed to the usual 10-15m between checks on this vehicle). That way, we can't get more than about a vehicle length into a trouble zone. And we make the slip checks pretty paranoid; if they fail to converge, we bail on the whole drive -- a common practice on Opportunity, but not something we usually bother with on Spirit. By the time we're done messing with it, we've got a drive that's likely to give up early for one reason or another. But better that, than be lost forever in El Dorado.
Mark Adler keeps discovering more about how things have changed since he was here last. "You're not doing anything during the 11 minutes of drive heating?" he asks incredulously. We used to pack the sol as full as possible; we'd have found something to do during those 11 minutes.
"It's not like it used to be," I shrug. "It grated on me for a while, but I got used to it. We have to make compromises to get out of here in shorter amounts of time, and to keep people from burning out." Sad, but true. I'd been thinking about just this when walking in this morning. When we changed the mission from a sprint to a marathon, we adjusted our pace accordingly. It was the right choice, and yet I regret being unable to use the vehicles to their fullest capacity -- it seems such a waste of a priceless scientific resource that the limiting factor in their operation would be us humans. I suppose that's a lesson to take forward to MSL: engineer them (and the ground tools, and whatever) so that they can be used to their fullest capacity in an eight-hour planning cycle.
But Mark remains stubbornly old-school in another way. "I'm gonna play a wakeup song," he announces. "It's 'El Dorado,' by ELO." He plays it for us. I'm sorry to say it's a pretty lousy song. But it's nice to have the old days back, even in such a small way, all the same.
[Next post: sol 710, January 1.]
2010-12-28
2010-12-21
Spirit Sol 699
Yestersol's IDD sequence had mixed results. The RAT failed to establish solid contact with the bumpy surface of this rock, so we didn't do the brush. Instead, the sequence simply proceeded to take MIs of the unbrushed surface and place the MB there.
This isn't the worst thing that could have happened. The MB is pretty good at seeing through the dust anyway, so we'll still get good science results from it even though the brushing never took place. However, the APXS can't see through the dust, so we really need to brush the darn thing thisol. We're falling farther and farther behind our drive metric all the time, and with Christmas coming this weekend, a failure to brush now could end up blowing most of a week. And that really wouldn't be good.
What happened yesterday was complex: almost exactly at the nominal contact position, the RAT's contact switches tripped briefly, but then rapidly gained and lost contact several times in less than a second. The initial firing of the contact switches stopped any further motion toward the rock face, but the sequence's subsequent check for contact failed because the switches had changed state by the time that check happened.
Clearly, we need some approach that checks whether we've made contact and autonomously tries again if it hasn't. The solution Frank and Matt Heverly and I come up with is something we've never tried before; we call it the "phased approach." First we'll try to place the RAT the same as we did yesterday, only without attempting any overdrive. (This will probably work, since we contacted 2mm early yestersol.) Then we'll have the sequence check for contact, and if it doesn't see contact, it'll push forward 5mm and check again. If still not in contact, it'll push forward 5mm and try again, then yet again, for a total of three tries (and 1.5cm total overdrive, about the most we dare command). And if we're still not in contact after all that, we'll try the whole routine again on an alternate target -- the same position, but with a slight tweak to the surface normal, so that we'll come in at a different angle.
If we're still not in contact, well ... then we give up. But at least we will have tried, and tried hard.
Though this approach sounds pretty complex, it's reasonably simple to sequence. I'm a good chunk of the way through it when Ashitey arrives to object. At first he's not comfortable brushing at all, since this switch-bouncing behavior isn't something we've seen before. He ultimately relents there, but still has problems with what we're up to.
"I think what you're doing is too complicated," he says. "You're doing things we've never done before." He doesn't say so outright, but this is clearly a sort of threat -- if we want to do what we're doing, he's going to make enough of a stink that we'll have to run it by Project management. "You should keep it simple -- just try again one time, with the tweaked surface normal."
"Besides," he continues, "you know it'll make contact if you just try again with the tweaked target."
But I don't know that. And I don't really think what we're doing is unreasonably complex, especially considering the downside: we have to stay here until we get this done, and since Home Plate is slipping farther and farther away every day, we should make every effort we can to get this to work (consistent with the safety of the vehicle, as always). And while it's true that we're doing something we've never done before, all of the parts of it are things that we have done before. "We invent techniques all the time without taking them to the Project level," I point out. "You think we should involve Jim Erickson every time we add something to a slip check?"
Ashitey backs off on the "first-time activity" line, but still pushes to make the thing simpler. We argue about it for a while, and perversely, our eventual compromise is even more complex than what we had when he came in. Now, we're first going to try the usual-style (non-phased) approach on the tweaked target. If that fails to make contact, we go to the original plan: try a phased approach on the original target, and if that fails to work, we try a phased approach on the tweaked target -- and only then, only if all that fails, we give up.
I'm not sure how that happened, but that's where we wind up.
All of this arguing pushes us pretty late -- by the time we're done, it's approaching 08:00, rover time, only an hour or so before we uplink. Someone remarks on this, but it fails to impress Mark Adler. "Ah, that's nothing," he says. "We used to CAM during the uplink window."
Well, I know they did that once. But it's not something anyone wants to repeat, you may be sure.
[Next post: sol 706, December 28.]
This isn't the worst thing that could have happened. The MB is pretty good at seeing through the dust anyway, so we'll still get good science results from it even though the brushing never took place. However, the APXS can't see through the dust, so we really need to brush the darn thing thisol. We're falling farther and farther behind our drive metric all the time, and with Christmas coming this weekend, a failure to brush now could end up blowing most of a week. And that really wouldn't be good.
What happened yesterday was complex: almost exactly at the nominal contact position, the RAT's contact switches tripped briefly, but then rapidly gained and lost contact several times in less than a second. The initial firing of the contact switches stopped any further motion toward the rock face, but the sequence's subsequent check for contact failed because the switches had changed state by the time that check happened.
Clearly, we need some approach that checks whether we've made contact and autonomously tries again if it hasn't. The solution Frank and Matt Heverly and I come up with is something we've never tried before; we call it the "phased approach." First we'll try to place the RAT the same as we did yesterday, only without attempting any overdrive. (This will probably work, since we contacted 2mm early yestersol.) Then we'll have the sequence check for contact, and if it doesn't see contact, it'll push forward 5mm and check again. If still not in contact, it'll push forward 5mm and try again, then yet again, for a total of three tries (and 1.5cm total overdrive, about the most we dare command). And if we're still not in contact after all that, we'll try the whole routine again on an alternate target -- the same position, but with a slight tweak to the surface normal, so that we'll come in at a different angle.
If we're still not in contact, well ... then we give up. But at least we will have tried, and tried hard.
Though this approach sounds pretty complex, it's reasonably simple to sequence. I'm a good chunk of the way through it when Ashitey arrives to object. At first he's not comfortable brushing at all, since this switch-bouncing behavior isn't something we've seen before. He ultimately relents there, but still has problems with what we're up to.
"I think what you're doing is too complicated," he says. "You're doing things we've never done before." He doesn't say so outright, but this is clearly a sort of threat -- if we want to do what we're doing, he's going to make enough of a stink that we'll have to run it by Project management. "You should keep it simple -- just try again one time, with the tweaked surface normal."
"Besides," he continues, "you know it'll make contact if you just try again with the tweaked target."
But I don't know that. And I don't really think what we're doing is unreasonably complex, especially considering the downside: we have to stay here until we get this done, and since Home Plate is slipping farther and farther away every day, we should make every effort we can to get this to work (consistent with the safety of the vehicle, as always). And while it's true that we're doing something we've never done before, all of the parts of it are things that we have done before. "We invent techniques all the time without taking them to the Project level," I point out. "You think we should involve Jim Erickson every time we add something to a slip check?"
Ashitey backs off on the "first-time activity" line, but still pushes to make the thing simpler. We argue about it for a while, and perversely, our eventual compromise is even more complex than what we had when he came in. Now, we're first going to try the usual-style (non-phased) approach on the tweaked target. If that fails to make contact, we go to the original plan: try a phased approach on the original target, and if that fails to work, we try a phased approach on the tweaked target -- and only then, only if all that fails, we give up.
I'm not sure how that happened, but that's where we wind up.
All of this arguing pushes us pretty late -- by the time we're done, it's approaching 08:00, rover time, only an hour or so before we uplink. Someone remarks on this, but it fails to impress Mark Adler. "Ah, that's nothing," he says. "We used to CAM during the uplink window."
Well, I know they did that once. But it's not something anyone wants to repeat, you may be sure.
[Next post: sol 706, December 28.]
2010-12-20
Spirit Sol 698
I love this rover. I really, really do.
I'm up at 05:45, and the first thing I do is log in and check the data from home. No images. That's just a touch worrisome. Maybe it's another OSS crash? But I see no evidence of that.
Hmmm ... I keep poking, and find that we did get some data -- so Spirit's alive -- but I find nothing to show that we've moved at all. It's as if the drive sequence never executed.
I shake it off, take a shower, and get to work as fast as I can. By the time I get there, the processing glitch -- if that's what it was -- has been resolved. We've got at least some drive data, and, most crucially, the post-drive images.
They show Comanche Spur dead in the middle of the IDD work volume. It's beautiful. We couldn't have gotten it more perfect. Or let me say, she couldn't have gotten it more perfect. All credit to our awesome rover.[1]
Visodom failed to converge at the end, and as a result, we didn't do our normal wheel wiggle, which we normally strive to do in order to settle the rover before deploying the IDD. When we look carefully at the images, though, we're able to convince ourselves that all wheels are in a stable configuration. This, coupled with past data and recent testbed work that has helped demonstrate that this rover design is highly stable when deploying the IDD even on apparently unstable surfaces, makes us willing to proceed.
What we proceed with is our pretty much standard IDD campaign -- MI, RAT-brush, MI, MB. Depending on the MB results we see tomorrow, we might stay here an extra sol, or cut things short and light out for our next destination, a large dark blob nicknamed "El Dorado."
Chris is RP-1 thisol, back on the rover after a FSW-induced absence of a couple of weeks, and he hasn't lost his touch. He cranks out the IDD sequence, I patch up a couple of minor errors, and the day's over before you can say "Jack Robinson."
Well, maybe not that fast. But pretty fast.
Courtesy NASA/JPL-Caltech. Comanche Spur, dead in our final front HAZCAM. I am more proud of this image than of almost any other in our nearly seven years on the surface.
Footnotes:
[1] I am unable to express just how proud I am of this drive to this very day. It was an amazing piece of work under extraordinary circumstances; if not my very best ever, it has to be damn close. Also, trying it was plain crazy, and hot damn, did it ever go perfectly. Big. Fat. Win.
I'm up at 05:45, and the first thing I do is log in and check the data from home. No images. That's just a touch worrisome. Maybe it's another OSS crash? But I see no evidence of that.
Hmmm ... I keep poking, and find that we did get some data -- so Spirit's alive -- but I find nothing to show that we've moved at all. It's as if the drive sequence never executed.
I shake it off, take a shower, and get to work as fast as I can. By the time I get there, the processing glitch -- if that's what it was -- has been resolved. We've got at least some drive data, and, most crucially, the post-drive images.
They show Comanche Spur dead in the middle of the IDD work volume. It's beautiful. We couldn't have gotten it more perfect. Or let me say, she couldn't have gotten it more perfect. All credit to our awesome rover.[1]
Visodom failed to converge at the end, and as a result, we didn't do our normal wheel wiggle, which we normally strive to do in order to settle the rover before deploying the IDD. When we look carefully at the images, though, we're able to convince ourselves that all wheels are in a stable configuration. This, coupled with past data and recent testbed work that has helped demonstrate that this rover design is highly stable when deploying the IDD even on apparently unstable surfaces, makes us willing to proceed.
What we proceed with is our pretty much standard IDD campaign -- MI, RAT-brush, MI, MB. Depending on the MB results we see tomorrow, we might stay here an extra sol, or cut things short and light out for our next destination, a large dark blob nicknamed "El Dorado."
Chris is RP-1 thisol, back on the rover after a FSW-induced absence of a couple of weeks, and he hasn't lost his touch. He cranks out the IDD sequence, I patch up a couple of minor errors, and the day's over before you can say "Jack Robinson."
Well, maybe not that fast. But pretty fast.
Courtesy NASA/JPL-Caltech. Comanche Spur, dead in our final front HAZCAM. I am more proud of this image than of almost any other in our nearly seven years on the surface.
Footnotes:
[1] I am unable to express just how proud I am of this drive to this very day. It was an amazing piece of work under extraordinary circumstances; if not my very best ever, it has to be damn close. Also, trying it was plain crazy, and hot damn, did it ever go perfectly. Big. Fat. Win.
2010-12-19
Spirit Sol 697
Amazingly, the drive went absolutely perfectly. During sequencing, I'd realized that if we changed the direction we turned when scuffing, we should be able to get a perfect view of Comanche in the FHAZ -- and there it is, large as life. The NCAM images are spectacular, showing just what we expected -- the larger, main part of Comanche on the left, a smaller spur on the right, and a kind of saddle in between.
The scuff, too, is right where it's supposed to be, although the soil here was hard enough that the scuff (Squyres derides it as a "smudge") isn't scientifically interesting. As a result, they decide after some discussion to bag IDDing the scuff and just do a couple of sols of remote sensing followed by an approach to Comanche.
So that's all good. But the OSS -- our shared filesystem -- crashed over the weekend, which means that all the automated processing of downlinked data is screwed up and backlogged, which in turn delays our start. And, of course, we're still in tight sols, and we're trying to do a three-sol plan today. All in all, it's a perfect day for a critical piece of infrastructure to have hiccups.
Fortunately, our drive is going to be perfectly simple. The most obvious approach, a Hillary-style drive partway up onto Comanche, won't work; the exposed rock face is too far up, and we'd have to cross too much loose sand to get there. But when we cast around, we find another solution, a couple of low, flat rocks lying on the ground on the northeast part of Comanche. They're only about 10m away, and we've got a nice, smooth path straight to 'em. Admittedly, they're dusty-looking, except for one small, possibly more-or-less vertical face, but them's the breaks. So that's nice and easy.
Only one fly in that ointment: when we show Steve where we're thinking of heading, he sums up his reaction in these words: "I hate it."
Oooo ... kaaaaayyyyy ....
Well, there is one other option we'd previously dismissed. We could just climb up into the saddle between the two chunks of rock; plenty of good stuff up there. This is more difficult, and somewhat more dangerous. Not to mention that thisol's RP-2 is Frank -- normally an Opportunist, not a Spirit -- which makes the whole thing a shade more risky.
We do it anyway. If all goes well, Spirit will climb more than a meter up a rocky 14-degree slope, hang a right, and drive across a relatively narrow channel to the spur. To her right will be a sharp sandy slope, which we might or might not get stuck on; to the left, a wide sandy patch, which we could bog down in. Monday, we'll be about 8m from here, with Comanche Spur dead in the center of our IDD work volume.
That, or we'll never hear from Spirit again, and have to live with ourselves forever.
Oddly enough, even though part of me says this is insane, I have so much history with this rover, and so much trust in her, that I'm not as worried about it as I'd expect. Am I right to think so? Monday morning, I'll know for sure.
Courtesy NASA/JPL-Caltech. Our final front HAZCAM, showing Comanche right in front of us.
Courtesy NASA/JPL-Caltech. A NAVCAM view of the same scene, aimed at the "saddle."
The scuff, too, is right where it's supposed to be, although the soil here was hard enough that the scuff (Squyres derides it as a "smudge") isn't scientifically interesting. As a result, they decide after some discussion to bag IDDing the scuff and just do a couple of sols of remote sensing followed by an approach to Comanche.
So that's all good. But the OSS -- our shared filesystem -- crashed over the weekend, which means that all the automated processing of downlinked data is screwed up and backlogged, which in turn delays our start. And, of course, we're still in tight sols, and we're trying to do a three-sol plan today. All in all, it's a perfect day for a critical piece of infrastructure to have hiccups.
Fortunately, our drive is going to be perfectly simple. The most obvious approach, a Hillary-style drive partway up onto Comanche, won't work; the exposed rock face is too far up, and we'd have to cross too much loose sand to get there. But when we cast around, we find another solution, a couple of low, flat rocks lying on the ground on the northeast part of Comanche. They're only about 10m away, and we've got a nice, smooth path straight to 'em. Admittedly, they're dusty-looking, except for one small, possibly more-or-less vertical face, but them's the breaks. So that's nice and easy.
Only one fly in that ointment: when we show Steve where we're thinking of heading, he sums up his reaction in these words: "I hate it."
Oooo ... kaaaaayyyyy ....
Well, there is one other option we'd previously dismissed. We could just climb up into the saddle between the two chunks of rock; plenty of good stuff up there. This is more difficult, and somewhat more dangerous. Not to mention that thisol's RP-2 is Frank -- normally an Opportunist, not a Spirit -- which makes the whole thing a shade more risky.
We do it anyway. If all goes well, Spirit will climb more than a meter up a rocky 14-degree slope, hang a right, and drive across a relatively narrow channel to the spur. To her right will be a sharp sandy slope, which we might or might not get stuck on; to the left, a wide sandy patch, which we could bog down in. Monday, we'll be about 8m from here, with Comanche Spur dead in the center of our IDD work volume.
That, or we'll never hear from Spirit again, and have to live with ourselves forever.
Oddly enough, even though part of me says this is insane, I have so much history with this rover, and so much trust in her, that I'm not as worried about it as I'd expect. Am I right to think so? Monday morning, I'll know for sure.
Courtesy NASA/JPL-Caltech. Our final front HAZCAM, showing Comanche right in front of us.
Courtesy NASA/JPL-Caltech. A NAVCAM view of the same scene, aimed at the "saddle."
2010-12-15
Spirit Sol 694
In summary: a good drive, but not great. We made 30m of progress, carrying us all the way to the ridge. But Murphy's Law is interplanetary, and visodom failed to converge during Spirit's top-of-ridge slip check. Which, thanks to our increased paranoia, meant she stopped driving.
As it happens, she probably would have stopped driving shortly after that anyway, since there is a rough semicircle of autonav-unfriendly tilts just a couple of meters below us. However, it's something we feel comfortable blind-driving through.
In fact, as we look down the ridge and across the terrain separating us from Comanche, it looks like we could blind-drive across the whole thing. It's a total of about 40m to the base of Comanche, and there are hardly any obstacles at all. They've allocated over three hours for driving again thisol, and I end up giving most of it back.
(The scientists have no problem figuring out what to do with the time, though. Turns out we're parked on the largest expanse of outcrop they've yet seen at Gusev, so they pack in plenty of PCAM and MTES imaging of it before we leave it behind forever. We'll also be taking a couple of pictures and relaying them through MEX as a friendly bit of international cooperation. Since our European friends plan to show them off, as one person notes, "Let's make 'em pretty." We settle on a nice color PCAM of Comanche and an NCAM of the intervening terrain. Should look pretty darn cool.)
The nice straight path makes the sequencing simple as well, which leaves me time to get lunch. As I'm walking over there with our prodigal son, Mark Adler, I ask him what impression he has of how MER's changed since he left.
"There's a lot of automation," he says immediately. "It all runs really smoothly. 'Course, if something goes wrong, well, that's a runout sol."
He's right about the increased automation, but as to the rest, I don't remember the last time an automation failure led to dropping the plan and executing the runout. I don't know if it's ever happened. We don't have the kind of team that likes to let it happen.
Another kind of automation failure gets us worried, though -- at one point, the file server freezes for a few minutes. We can't do anything without it, so this brings the process to a grinding halt. Fortunately, it comes back after a short delay, and the ball resumes rolling.
While all this is going on in the Spirit World, Opportunity's been making progress of her own. Slowly but surely, they're performing a limited IDD campaign on the juicy bit of outcrop we parked in front of, lo, these many sols ago. In the process, they're learning about the fault, and one of the things they learned yesterday is that the amount of resistance caused by the broken wire isn't a constant. Instead, it seems to vary as the arm articulates.
But the way they found out was by having another joint stall, so they're in the middle of a tool change -- with the MI pointing straight up and its dust cover open, to make matters worse. It was supposed to be a skip sol, but they need someone to help them recover and get things back on track. And who just became available, but little old me?
So I go upstairs and help Ashitey sequence the recovery tool change. It's not quite the first time a rover driver has worked on both rovers in one sol -- if nothing else, Cooper did it when the rest of us were in Hawaii -- but it's the first time I've done it. ("A personal best," as Emily Eelkema puts it.)
Two Mars rovers in one day, baby. It's every geek's dream.
[Next post: sol 697, December 19.]
As it happens, she probably would have stopped driving shortly after that anyway, since there is a rough semicircle of autonav-unfriendly tilts just a couple of meters below us. However, it's something we feel comfortable blind-driving through.
In fact, as we look down the ridge and across the terrain separating us from Comanche, it looks like we could blind-drive across the whole thing. It's a total of about 40m to the base of Comanche, and there are hardly any obstacles at all. They've allocated over three hours for driving again thisol, and I end up giving most of it back.
(The scientists have no problem figuring out what to do with the time, though. Turns out we're parked on the largest expanse of outcrop they've yet seen at Gusev, so they pack in plenty of PCAM and MTES imaging of it before we leave it behind forever. We'll also be taking a couple of pictures and relaying them through MEX as a friendly bit of international cooperation. Since our European friends plan to show them off, as one person notes, "Let's make 'em pretty." We settle on a nice color PCAM of Comanche and an NCAM of the intervening terrain. Should look pretty darn cool.)
The nice straight path makes the sequencing simple as well, which leaves me time to get lunch. As I'm walking over there with our prodigal son, Mark Adler, I ask him what impression he has of how MER's changed since he left.
"There's a lot of automation," he says immediately. "It all runs really smoothly. 'Course, if something goes wrong, well, that's a runout sol."
He's right about the increased automation, but as to the rest, I don't remember the last time an automation failure led to dropping the plan and executing the runout. I don't know if it's ever happened. We don't have the kind of team that likes to let it happen.
Another kind of automation failure gets us worried, though -- at one point, the file server freezes for a few minutes. We can't do anything without it, so this brings the process to a grinding halt. Fortunately, it comes back after a short delay, and the ball resumes rolling.
While all this is going on in the Spirit World, Opportunity's been making progress of her own. Slowly but surely, they're performing a limited IDD campaign on the juicy bit of outcrop we parked in front of, lo, these many sols ago. In the process, they're learning about the fault, and one of the things they learned yesterday is that the amount of resistance caused by the broken wire isn't a constant. Instead, it seems to vary as the arm articulates.
But the way they found out was by having another joint stall, so they're in the middle of a tool change -- with the MI pointing straight up and its dust cover open, to make matters worse. It was supposed to be a skip sol, but they need someone to help them recover and get things back on track. And who just became available, but little old me?
So I go upstairs and help Ashitey sequence the recovery tool change. It's not quite the first time a rover driver has worked on both rovers in one sol -- if nothing else, Cooper did it when the rest of us were in Hawaii -- but it's the first time I've done it. ("A personal best," as Emily Eelkema puts it.)
Two Mars rovers in one day, baby. It's every geek's dream.
[Next post: sol 697, December 19.]
2010-12-14
Spirit Sol 693
We made a respectable 25m of progress -- good, but not great. The visodom and blind segments went fine, but autonav failed shortly afterward, after finding no safe path. Oddly, we're not sure why autonav didn't want to proceed; what we can see in the images looks good.
Anyway, we're slowly but surely gaining on Comanche. We've still got about 35m to the ridge, and we've got a good chance of getting there and beyond because they've scared up over three hours of drive time today.
The path from here will be an odd one. After a bit of maneuvering to get onto the mesh, we've got to make our way through a trio of rocks arranged as the vertices of a roughly equilateral triangle, whose closest side lies transverse to our path about 10m from here. Our plan is to drive between the rocks to the center of the triangle, then jog left to dodge the rock at its apex before getting back on course and heading to the ridge.
About 10m after that -- still before we get to the ridge -- there's enough undulation in the terrain that our mesh starts to drop out. So we'll switch on autonav for the remaining 12m to the ridge, then turn and head to Comanche.
We're still worried about that ridge, though. I think there's a better-than-even chance that the far side will be too steep or too rocky for autonav to make any progress, in which case I'll feel guilty tomorrow about blowing a lot of drive time. But there's not too much I can do about that and still keep the rover safe.
One other modification we make to our usual Spirit-driving practices is to make the slip check more paranoid. Before starting autonav, and periodically during autonav drives, we've been having Spirit perform slip checks, looking to see whether she's slipping too much and stopping the drive early if so. Normally, if visodom fails to converge -- that is, if the slip-check calculations don't work, so we don't know what our slip is -- we let her drive on anyway. After all, that's a fairly unusual case, and even if that slip check fails, there will be another one shortly.
But in this case, we're just nervous about that. Once she gets over that ridge, she could be in really bad territory, and if she can't prove she's safe to go on, we want her to stop. So we tell her to do that. Will it cut our day short? We'll find out tomorrow ... same Mars time, same Mars channel.
Courtesy NASA/JPL-Caltech. Comanche, dead ahead.
Courtesy NASA/JPL-Caltech. In the other direction, a look back at our tracks. My girl's such a trouper.
Anyway, we're slowly but surely gaining on Comanche. We've still got about 35m to the ridge, and we've got a good chance of getting there and beyond because they've scared up over three hours of drive time today.
The path from here will be an odd one. After a bit of maneuvering to get onto the mesh, we've got to make our way through a trio of rocks arranged as the vertices of a roughly equilateral triangle, whose closest side lies transverse to our path about 10m from here. Our plan is to drive between the rocks to the center of the triangle, then jog left to dodge the rock at its apex before getting back on course and heading to the ridge.
About 10m after that -- still before we get to the ridge -- there's enough undulation in the terrain that our mesh starts to drop out. So we'll switch on autonav for the remaining 12m to the ridge, then turn and head to Comanche.
We're still worried about that ridge, though. I think there's a better-than-even chance that the far side will be too steep or too rocky for autonav to make any progress, in which case I'll feel guilty tomorrow about blowing a lot of drive time. But there's not too much I can do about that and still keep the rover safe.
One other modification we make to our usual Spirit-driving practices is to make the slip check more paranoid. Before starting autonav, and periodically during autonav drives, we've been having Spirit perform slip checks, looking to see whether she's slipping too much and stopping the drive early if so. Normally, if visodom fails to converge -- that is, if the slip-check calculations don't work, so we don't know what our slip is -- we let her drive on anyway. After all, that's a fairly unusual case, and even if that slip check fails, there will be another one shortly.
But in this case, we're just nervous about that. Once she gets over that ridge, she could be in really bad territory, and if she can't prove she's safe to go on, we want her to stop. So we tell her to do that. Will it cut our day short? We'll find out tomorrow ... same Mars time, same Mars channel.
Courtesy NASA/JPL-Caltech. Comanche, dead ahead.
Courtesy NASA/JPL-Caltech. In the other direction, a look back at our tracks. My girl's such a trouper.
2010-12-13
Spirit Sol 692
The drive went well. Spirit hit a limit cycle check -- she wasn't making progress fast enough, so she stopped driving after the first autonav waypoint -- but she still made a respectable 30m of progress. We're far enough around Miami (which looks even more impressive from here, as we've descended about 3m) to see Comanche from the other side.
And the news from here is not so bad. Rather than a shallow ramp all the way to the valley floor, it looks like we've still got a ridge to cross, but we've got a decent path to it, at least. And no time pressure -- today's Spirit's skip day, so Frank and Jeng and Ashley and I are planning the drive a day ahead. Frank and Jeng are as experienced at rover driving as they come, but they spend all their time on Opportunity, so we're avoiding making them plan a drive on an unfamiliar rover on a tight schedule, which is what would happen if they waited until tomorrow.
It's a good opportunity for me to revisit what's different between the vehicles. Frank actually has to ask what's the height of obstacles we're worried about -- after all, they never see anything that big on the other side of the world. "When we see a rock over there, they want us to drive onto it," he points out. And he's quite right: for months, the only rocks Opportunity's seen have been either tiny cobbles or, more commonly, flat outcrop patches that make for good driving.
Well, they showed up for an interesting drive. We've got several actual obstacles to avoid, so we'll have a visodom segment at the beginning to help navigate around them, then a short blind drive through a friendly patch, then autonav from there. I don't think we'll quite make it, but if we get really lucky, we'll make it to the ridge and have a spectacular view of the valley beyond. (And an equally good view of whatever we're going to have to cross to get down there.) If that doesn't happen this drive, it'll happen on the next one. And since I'm working on Spirit for the rest of the year, while they continue to investigate Opportunity's IDD problem, I'll be sure to see it.
I can't wait!
And the news from here is not so bad. Rather than a shallow ramp all the way to the valley floor, it looks like we've still got a ridge to cross, but we've got a decent path to it, at least. And no time pressure -- today's Spirit's skip day, so Frank and Jeng and Ashley and I are planning the drive a day ahead. Frank and Jeng are as experienced at rover driving as they come, but they spend all their time on Opportunity, so we're avoiding making them plan a drive on an unfamiliar rover on a tight schedule, which is what would happen if they waited until tomorrow.
It's a good opportunity for me to revisit what's different between the vehicles. Frank actually has to ask what's the height of obstacles we're worried about -- after all, they never see anything that big on the other side of the world. "When we see a rock over there, they want us to drive onto it," he points out. And he's quite right: for months, the only rocks Opportunity's seen have been either tiny cobbles or, more commonly, flat outcrop patches that make for good driving.
Well, they showed up for an interesting drive. We've got several actual obstacles to avoid, so we'll have a visodom segment at the beginning to help navigate around them, then a short blind drive through a friendly patch, then autonav from there. I don't think we'll quite make it, but if we get really lucky, we'll make it to the ridge and have a spectacular view of the valley beyond. (And an equally good view of whatever we're going to have to cross to get down there.) If that doesn't happen this drive, it'll happen on the next one. And since I'm working on Spirit for the rest of the year, while they continue to investigate Opportunity's IDD problem, I'll be sure to see it.
I can't wait!
2010-12-11
Spirit Sol 690
We didn't get all the data we were supposed to get -- it's stuck someplace on Earth, but hasn't made it to the flight ops LAN yet.
The question is whether we can still do the drive with what we've got. (Or rather, with what we had yesterday: we came in then and planned most of the drive already, since there was no time pressure that way.)
We could do a better drive if we had all the data, but it looks like we've got enough -- at least we have NAVCAM coverage of the areas to the right and left of Miami, which towers over us, like, uh, a tower or something. What we can see of the direct path to Comanche is unpromising; most paths are blocked or worrisomely bumpy, and from what we can see, the one path we find that leads straight to the ridge looks like we'd more or less fall right over it. Visions of Wile E. Coyote dance in my head. Of course, we'd have the rover on autonav, so she'd keep herself safe, but our best estimate, backed up by what we can see of this area from images taken higher on the hill, is that the path down from there is steep and bumpy.
This contrasts with the longer path that leads to the left (east) around Miami -- which I take to calling the Miami Bypass. There the route is relatively smooth along the area we can see, and the earlier images suggest a smooth and gentle path. A longer one, but an easier one; all things considered, the long way might be faster. So we go that way.
It's starting to feel like Old Timer's Day around here. Mark Adler's been here recently, though not today. But Julie Townsend is back, as is Matt Keuneke, and Rob Manning stops by for a visit. As does Steve Squyres, which is always a pleasant surprise. He's on a layover between flights or something, and couldn't stay away.
He also bugs us about the drive we're planning -- why are we going to the left of Miami, not to the right? He's only seen the 2-D versions of the images, as it turns out. I bring up the true-stereo 3-D images in RSVP, and Steve slaps on the goggles. It just takes one glance. "I'm convinced!"
[Next post: sol 692, December 13.]
The question is whether we can still do the drive with what we've got. (Or rather, with what we had yesterday: we came in then and planned most of the drive already, since there was no time pressure that way.)
We could do a better drive if we had all the data, but it looks like we've got enough -- at least we have NAVCAM coverage of the areas to the right and left of Miami, which towers over us, like, uh, a tower or something. What we can see of the direct path to Comanche is unpromising; most paths are blocked or worrisomely bumpy, and from what we can see, the one path we find that leads straight to the ridge looks like we'd more or less fall right over it. Visions of Wile E. Coyote dance in my head. Of course, we'd have the rover on autonav, so she'd keep herself safe, but our best estimate, backed up by what we can see of this area from images taken higher on the hill, is that the path down from there is steep and bumpy.
This contrasts with the longer path that leads to the left (east) around Miami -- which I take to calling the Miami Bypass. There the route is relatively smooth along the area we can see, and the earlier images suggest a smooth and gentle path. A longer one, but an easier one; all things considered, the long way might be faster. So we go that way.
It's starting to feel like Old Timer's Day around here. Mark Adler's been here recently, though not today. But Julie Townsend is back, as is Matt Keuneke, and Rob Manning stops by for a visit. As does Steve Squyres, which is always a pleasant surprise. He's on a layover between flights or something, and couldn't stay away.
He also bugs us about the drive we're planning -- why are we going to the left of Miami, not to the right? He's only seen the 2-D versions of the images, as it turns out. I bring up the true-stereo 3-D images in RSVP, and Steve slaps on the goggles. It just takes one glance. "I'm convinced!"
[Next post: sol 692, December 13.]
2010-12-08
Spirit Sol 687
Second time was the charm. "As usual, a perfect drive!" exclaims the SOWG chair. "As usual," John mutters sardonically, "not as always."
Well, be that as it may. They have indeed revised the drive metric, making us a whopping 100m behind schedule. So we're driving.
Ha, ha! Only kidding. Of course we're not driving! Instead, we're spending the next couple of sols IDDing the patch of outcrop this drive took us to. And, of course, imaging nearby rocks and such. The outcrop is named Algonquin, and the target names have been selected from a list of tongue-twisting Indian names. After listening to the team stumbling over the names in the SOWG room, I thumb the mike and say, "I just wanted to remind everybody that the reason we name features is to make them easier to refer to." This wins laughter and applause. The SOWG chair proposes that our next target namespace be limited to three-letter names.
The IDD sequencing is our usual abbreviated campaign: on the first sol, an MI mosaic followed by an APXS placement; on the second, a RAT brush followed by a repeat of the MI mosaic and APXS placement. John has this well in hand -- true to form, he mostly copies and pastes from a previous campaign -- which leaves me time to look ahead to Friday's drive.
Which doesn't look pretty. About 15m away is a ridge, beyond which we can see terrain, but we have no range data on it -- that is, we have no idea how far away it is. PCAMs usually can give us range data out to 100m or so, so the fact that we're getting nothing suggests that what we see is at least that much farther away. So our strategy must be to drive to the ridge, then switch on autonav and hope for the best beyond that.
If we can even get to the ridge. The terrain here is just rugged enough to make that really, really difficult -- every apparent solution is thwarted by some rock or other.
Moreover, the target beyond, a big red mound called Comanche, is part of the stuff we have no range data for. So we don't know how far away it is, and thus we don't know how long it'll take us to drive there.
Because Monday's such a tight sol, we don't want to plan a drive that sol. Instead, we'll take the unusual step of planning two drives Friday. The first will take us to the ridge, then autonav past it. The second will be an all-autonav drive. Since we're fairly sure Comanche is more than 100m away, and the two drives combined aren't likely to make that much distance, we can just pick a point ~ 100m away and have both drives aim for it.
We've never done this on Spirit, but it's similar to what we did on Opportunity over President's Day weekend, so we're confident it'll work. We end up scrapping the idea, though, less for technical than for staffing reasons. Next week, John will be absent because of his fiancee Helen's surgery, Ashitey is still tied up with debugging Opportunity's IDD, and Chris Leger is focused on the FSW release. This leaves me as the only usual Spirit RP, and I have an unavoidable doctor's appointment Tuesday, a follow-up with my surgeon.
As a result, Sharon's had to schedule two Opportunity RPs, Frank and Jeng, for Tuesday. And we really don't want to make them drive. They're both good RPs, and everything would probably go fine, but the combination of a tight uplink, two RPs somewhat unfamiliar with Spirit, and unknown and likely hazardous terrain makes us nervous enough to call it off.
Instead, Friday we'll plan just one drive, and Monday we'll look at the data from that drive and plan the next one. We won't send that sequence Monday, though -- Monday's schedule is just too tight for that -- we'll hang onto it and have Frank and Jeng send it Tuesday.
This means I get to work on the rovers Monday, instead of spending Monday pushing on the RSVP software release for Phoenix. I can't say I'm disappointed.
[Next post: sol 690, December 11.]
Well, be that as it may. They have indeed revised the drive metric, making us a whopping 100m behind schedule. So we're driving.
Ha, ha! Only kidding. Of course we're not driving! Instead, we're spending the next couple of sols IDDing the patch of outcrop this drive took us to. And, of course, imaging nearby rocks and such. The outcrop is named Algonquin, and the target names have been selected from a list of tongue-twisting Indian names. After listening to the team stumbling over the names in the SOWG room, I thumb the mike and say, "I just wanted to remind everybody that the reason we name features is to make them easier to refer to." This wins laughter and applause. The SOWG chair proposes that our next target namespace be limited to three-letter names.
The IDD sequencing is our usual abbreviated campaign: on the first sol, an MI mosaic followed by an APXS placement; on the second, a RAT brush followed by a repeat of the MI mosaic and APXS placement. John has this well in hand -- true to form, he mostly copies and pastes from a previous campaign -- which leaves me time to look ahead to Friday's drive.
Which doesn't look pretty. About 15m away is a ridge, beyond which we can see terrain, but we have no range data on it -- that is, we have no idea how far away it is. PCAMs usually can give us range data out to 100m or so, so the fact that we're getting nothing suggests that what we see is at least that much farther away. So our strategy must be to drive to the ridge, then switch on autonav and hope for the best beyond that.
If we can even get to the ridge. The terrain here is just rugged enough to make that really, really difficult -- every apparent solution is thwarted by some rock or other.
Moreover, the target beyond, a big red mound called Comanche, is part of the stuff we have no range data for. So we don't know how far away it is, and thus we don't know how long it'll take us to drive there.
Because Monday's such a tight sol, we don't want to plan a drive that sol. Instead, we'll take the unusual step of planning two drives Friday. The first will take us to the ridge, then autonav past it. The second will be an all-autonav drive. Since we're fairly sure Comanche is more than 100m away, and the two drives combined aren't likely to make that much distance, we can just pick a point ~ 100m away and have both drives aim for it.
We've never done this on Spirit, but it's similar to what we did on Opportunity over President's Day weekend, so we're confident it'll work. We end up scrapping the idea, though, less for technical than for staffing reasons. Next week, John will be absent because of his fiancee Helen's surgery, Ashitey is still tied up with debugging Opportunity's IDD, and Chris Leger is focused on the FSW release. This leaves me as the only usual Spirit RP, and I have an unavoidable doctor's appointment Tuesday, a follow-up with my surgeon.
As a result, Sharon's had to schedule two Opportunity RPs, Frank and Jeng, for Tuesday. And we really don't want to make them drive. They're both good RPs, and everything would probably go fine, but the combination of a tight uplink, two RPs somewhat unfamiliar with Spirit, and unknown and likely hazardous terrain makes us nervous enough to call it off.
Instead, Friday we'll plan just one drive, and Monday we'll look at the data from that drive and plan the next one. We won't send that sequence Monday, though -- Monday's schedule is just too tight for that -- we'll hang onto it and have Frank and Jeng send it Tuesday.
This means I get to work on the rovers Monday, instead of spending Monday pushing on the RSVP software release for Phoenix. I can't say I'm disappointed.
[Next post: sol 690, December 11.]
2010-12-06
Spirit Sol 685
The weekend drive faulted out right at the start, before even going anywhere. Painfully, it was really a rookie mistake. They fired off a helper sequence, and, as we usually do, put in a check to see if the helper was present, precluding further driving if not. The problem was, they accidentally reversed the sense of the test, precluding driving if the helper was present. Which of course it was, so the drive kicked off the helper sequence and then refused to go anywhere.
I'm just glad it wasn't me.[1]
Partly as a result of this mistake, we're now well below our drive metric. Also contributing to the problem is that they're going to raise the bar. Recognizing that our power has dropped and will continue to drop, we're revising our estimate of when we need to make it to Home Plate. The result will be more driving -- and more pressure on us to get it right.
Not that they seem to be in a terrible hurry. This drive will be all of 15m or so, to a nearby rock target called Algonquin. Where we'll stop to IDD. "But let's have the discipline," SOWG chair Albert Yen cautions, "to spend only one drive sol there." Since we're in restricted sols, that really means spending two drive sols there -- we'll drive only once more this week.
This mistake is the third time in about a week that we've blown all or part of a drive. The first case was when we hit the tilt limit right at the start of an autonav drive, which cut that sol from something like 40m to something like 20m. The second was when we screwed up the time-of-day limit logic, blowing our turn for comm. (At least that one made all of its planned distance, but the mistake is still troubling.) And now this is the third one.
I talk about these mistakes with John and Ashley, to see if we can come up with fundamental underlying reasons. "Complacency" is John's answer, which might be as true as anything. Still, we come up with a couple of new flight rules, which I write automated checks for. Maybe they'll help.
Later, I talk to Frank about how Opportunity is doing. There's good news and bad news, he tells me. The bad news is that we're still not seeing any motion from Opportunity's IDD. The good news is that Joe Melko has identified a failure mode in which a wire cracks, doubling the resistance. And there seems to be some support for this theory in the data. If this were true, we'd be able to resume normal work by just doubling the current through that joint.
If it's true. I can't help getting my hopes up, but I have to remind myself that we still don't know what's wrong, and the IDD could still be lost forever.
I'm hoping not.
[Next post: sol 687, December 8.]
Footnotes:
[1] Technically, I'm glad it wasn't I.
I'm just glad it wasn't me.[1]
Partly as a result of this mistake, we're now well below our drive metric. Also contributing to the problem is that they're going to raise the bar. Recognizing that our power has dropped and will continue to drop, we're revising our estimate of when we need to make it to Home Plate. The result will be more driving -- and more pressure on us to get it right.
Not that they seem to be in a terrible hurry. This drive will be all of 15m or so, to a nearby rock target called Algonquin. Where we'll stop to IDD. "But let's have the discipline," SOWG chair Albert Yen cautions, "to spend only one drive sol there." Since we're in restricted sols, that really means spending two drive sols there -- we'll drive only once more this week.
This mistake is the third time in about a week that we've blown all or part of a drive. The first case was when we hit the tilt limit right at the start of an autonav drive, which cut that sol from something like 40m to something like 20m. The second was when we screwed up the time-of-day limit logic, blowing our turn for comm. (At least that one made all of its planned distance, but the mistake is still troubling.) And now this is the third one.
I talk about these mistakes with John and Ashley, to see if we can come up with fundamental underlying reasons. "Complacency" is John's answer, which might be as true as anything. Still, we come up with a couple of new flight rules, which I write automated checks for. Maybe they'll help.
Later, I talk to Frank about how Opportunity is doing. There's good news and bad news, he tells me. The bad news is that we're still not seeing any motion from Opportunity's IDD. The good news is that Joe Melko has identified a failure mode in which a wire cracks, doubling the resistance. And there seems to be some support for this theory in the data. If this were true, we'd be able to resume normal work by just doubling the current through that joint.
If it's true. I can't help getting my hopes up, but I have to remind myself that we still don't know what's wrong, and the IDD could still be lost forever.
I'm hoping not.
[Next post: sol 687, December 8.]
Footnotes:
[1] Technically, I'm glad it wasn't I.
2010-12-03
Opportunity Sol 660 (Spirit Sol 682)
I'm not needed to cover Ashitey's shift thisol, so I'm back on Opportunity, where things are not looking up. The day starts with a meeting in which the usual suspects are gathered around the table (or on the phone line) for a review of the spacecraft's state, and to try to figure out where we go from here.
In short: no joy. The IDD is showing small or no motion. To make matters worse, since it's in the stowed configuration, where there's very little room for it to move, we can't command large enough motions to get a lot of data. So there are few new facts to argue about, which is good or bad depending on how you look at it. Mostly bad.
Thisol will be more of the same: a couple of short motions designed to rotate the arm slightly -- just a couple of degrees -- forward, the usual next step in the unstow process. We're going to be a little more liberal in what we allow, permitting the startup current to be high for a little longer, in the hope that this will get it over whatever hump it needs to get over in order to move. And we're taking a bunch of pictures. And keeping our fingers crossed.
[Next post: sol 685, December 6.]
In short: no joy. The IDD is showing small or no motion. To make matters worse, since it's in the stowed configuration, where there's very little room for it to move, we can't command large enough motions to get a lot of data. So there are few new facts to argue about, which is good or bad depending on how you look at it. Mostly bad.
Thisol will be more of the same: a couple of short motions designed to rotate the arm slightly -- just a couple of degrees -- forward, the usual next step in the unstow process. We're going to be a little more liberal in what we allow, permitting the startup current to be high for a little longer, in the hope that this will get it over whatever hump it needs to get over in order to move. And we're taking a bunch of pictures. And keeping our fingers crossed.
[Next post: sol 685, December 6.]
2010-12-01
Spirit Sol 680
Our drive failed. Well, that might be too strong a word. We got through the blind portion, the initial 25m or so, but when we switched to autonav, we lowered the tilt limit (as usual), almost immediately drove over a relatively sharp little ramp, and tilt-faulted.
What makes this more annoying is that, unbeknownst to me, we had terrain meshes covering that entire area -- from an earlier sol, sol 667, when they'd driven up to that point and looked out over it. (I was on Opportunity then and missed that part of the story. When I came in, they'd driven back to examine some outcrop they'd passed by previously.) So we could easily have done a blind drive right through this area -- with a higher tilt limit, so we wouldn't have faulted.
Foo.
Just to make things more awkward, there was a problem at the DSN station, and they somehow lost a bunch of data from our downlink. (The data somehow made it all the way from Mars to Spain, but couldn't quite make it the last few thousand km to California. This is like swimming the Atlantic and drowning a meter from shore.) As a result, we have very little imagery from our current position -- HAZCAMs are all we had last night, though the PCAMs showed up in the morning pass. We're still missing the creamy middle -- I mean, crucial middle -- that's ordinarily supplied by the NCAMs.
We'll be able to do a drive, but it will be mostly autonav and therefore rather short -- but, hey. We still have that imagery from sol 667! John loads up the terrain mesh, and though it shows we've accumulated significant position error, we've still got good enough data to extend the blind drive to 12m or so. That gets us to the start of the PCAM coverage, which would ordinarily mean we'd be able to extend the blind drive even further, but there's some scary-looking stuff at the start of the PCAM coverage that we don't want to blind-drive through. Oh, well.
Anyway, there's a growing sentiment for bypassing Miami in favor of Comanche, a slightly more distant target that's redder (as opposed to Miami's black, which might suggest it's just regular old basalt) and has a more unusual morphology. So we aim this drive for a spot to the right of Miami -- somewhat splitting the difference between it and Comanche. And off we go.
Just after I deliver, I check the data -- and sure enough, the NCAMs have just come down. And they show what we suspected even without them: we'd have been perfectly fine sequencing this entire drive as a blind drive, which would have more than doubled our distance. Too late to change it now. Double-foo.
While I've been covering Ashitey's shifts on this side of the planet, he's been on the other side of the planet, investigating Opportunity's IDD anomaly. And the news isn't good, or at least, that's what Frank tells me. They haven't done anything with Opportunity's IDD since the fault, but they've been busy in the testbed, and their testing there has tended to disconfirm the hypothesis that the IDD is simply temporarily having trouble getting off its hook. Rather, it's looking more and more like a real actuator problem, a hardware problem we won't be able to solve and may not be able to work around.
I'll be back on Opportunity tomorrow morning, starting with an early meeting to explore the results of the tiny move they'll send up tonight. And then we'll know more.
[Next post: sol 682 (Opportunity sol 660), December 3.]
What makes this more annoying is that, unbeknownst to me, we had terrain meshes covering that entire area -- from an earlier sol, sol 667, when they'd driven up to that point and looked out over it. (I was on Opportunity then and missed that part of the story. When I came in, they'd driven back to examine some outcrop they'd passed by previously.) So we could easily have done a blind drive right through this area -- with a higher tilt limit, so we wouldn't have faulted.
Foo.
Just to make things more awkward, there was a problem at the DSN station, and they somehow lost a bunch of data from our downlink. (The data somehow made it all the way from Mars to Spain, but couldn't quite make it the last few thousand km to California. This is like swimming the Atlantic and drowning a meter from shore.) As a result, we have very little imagery from our current position -- HAZCAMs are all we had last night, though the PCAMs showed up in the morning pass. We're still missing the creamy middle -- I mean, crucial middle -- that's ordinarily supplied by the NCAMs.
We'll be able to do a drive, but it will be mostly autonav and therefore rather short -- but, hey. We still have that imagery from sol 667! John loads up the terrain mesh, and though it shows we've accumulated significant position error, we've still got good enough data to extend the blind drive to 12m or so. That gets us to the start of the PCAM coverage, which would ordinarily mean we'd be able to extend the blind drive even further, but there's some scary-looking stuff at the start of the PCAM coverage that we don't want to blind-drive through. Oh, well.
Anyway, there's a growing sentiment for bypassing Miami in favor of Comanche, a slightly more distant target that's redder (as opposed to Miami's black, which might suggest it's just regular old basalt) and has a more unusual morphology. So we aim this drive for a spot to the right of Miami -- somewhat splitting the difference between it and Comanche. And off we go.
Just after I deliver, I check the data -- and sure enough, the NCAMs have just come down. And they show what we suspected even without them: we'd have been perfectly fine sequencing this entire drive as a blind drive, which would have more than doubled our distance. Too late to change it now. Double-foo.
While I've been covering Ashitey's shifts on this side of the planet, he's been on the other side of the planet, investigating Opportunity's IDD anomaly. And the news isn't good, or at least, that's what Frank tells me. They haven't done anything with Opportunity's IDD since the fault, but they've been busy in the testbed, and their testing there has tended to disconfirm the hypothesis that the IDD is simply temporarily having trouble getting off its hook. Rather, it's looking more and more like a real actuator problem, a hardware problem we won't be able to solve and may not be able to work around.
I'll be back on Opportunity tomorrow morning, starting with an early meeting to explore the results of the tiny move they'll send up tonight. And then we'll know more.
[Next post: sol 682 (Opportunity sol 660), December 3.]
2010-11-29
Spirit Sol 678
Opportunity experienced a disturbing failure this weekend: her IDD failed to deploy. We talked about it this weekend long enough to rule out a sequencing error, which meant we shouldn't try to bring in the normal uplink team to try to fix it and salvage the IDD work. It's possible we would have tried to get the right people together, but it was Thanksgiving, and many people were out of town, and so on. And these days we don't engage in heroics. So we let it go until Monday.
Now it's Monday, and it's not immediately clear what we should do about this, because it's not immediately clear what the failure is. We know we got a tiny amount of motion followed by a motor stall, but why? Eric Baumgartner's (well-educated) guess is that we've simply lost a small amount of knowledge as to the position of the IDD. As a result, when the IDD tried to go to the commanded position as part of its unstow routine, it actually failed to get the IDD off of the little hook it stows itself on, and then it couldn't rotate itself away from the hook. This hypothesis has some confirmation in some image analysis Paolo performed, which may show a tiny discrepancy in the arm's position after the last time we stowed it.
This would be relatively straightforward to solve -- not easy, but not dreadfully bad. We'd have to figure out where the arm actually is, then manually guide it through the unstow routine, at which point we'd be able to recalibrate the IDD by driving each of the joints against its hardstop. We might also need to repeat the "Martian Tai Chi" routine that recalibrates the camera models, so that the IDD and the cameras would be in agreement. It would be a long, slow process -- maybe a week or two, all put together -- but in the end we'd be as good as new.
But Eric's explanation isn't the only one that fits the observed facts. Ashitey fears there's something worse going on. One of the actuators in the testbed rover has started to fail, and it stalls out in pretty much the same way as we've just observed Opportunity's joint motor stall. If this is what's going on, it means the IDD has become unreliable -- every once in a while, our IDD work might randomly fail in this way. That would introduce a severe, somewhat random penalty into all IDD work.
With the testbed rover's IDD, the failures are merely annoying; repeat the failed command, and it usually works the second try. If Opportunity's IDD is indeed failing in this way, that means we have an occasional one-sol penalty to look forward to; probably nothing worse than that (though that would be unpleasant). But it could be worse still: the actuator might simply have failed permanently. And if that's what happened, a large fraction of Opportunity's scientific potential just disappeared forever.
There are other possible explanations, too, with varying degrees of benignity. Erickson wants the full-up anomaly investigation, and the right guy to do that is Ashitey. But Ashitey's RP-1 on Spirit today, which would leave him precious little time for something like that. So he and I switch places -- he heads up the investigation, and I go back to Spirit.
Which means I'm driving today -- driving to Miami. (I try to remember not to leave the left turn signal on.) This drive is reasonably simple, snaking around a few rocks before heading more or less straight toward Miami. Well, not quite Miami -- Coral Gables. See, I didn't have an exact target for Miami, just picked a spot in the neighborhood, and Coral Gables is a Miami suburb ....
Maybe you had to be there.
If all goes well, we'll cover a third to a half of the distance to Miami in this drive. (Maybe we'll stop for the night at a HoJo's. Sorry, I can't help it.)
I'd seen a birthday card in the Spirit sequencing room last week, but it hadn't registered on me until today. I pick it up and read it. I figured it was for someone on the team. Turns out I was right; it's for the most important member of our team. It's a birthday card for a one-year-old -- our one-year-old, Spirit. (Or maybe that should be a "birthsol" card, but I'm not picky.) From the nice (and clever) folks at unmannedspaceflight.com.
I love it. We should make them Spirit's godparents or something.
[Next post: sol 680, December 1.]
Now it's Monday, and it's not immediately clear what we should do about this, because it's not immediately clear what the failure is. We know we got a tiny amount of motion followed by a motor stall, but why? Eric Baumgartner's (well-educated) guess is that we've simply lost a small amount of knowledge as to the position of the IDD. As a result, when the IDD tried to go to the commanded position as part of its unstow routine, it actually failed to get the IDD off of the little hook it stows itself on, and then it couldn't rotate itself away from the hook. This hypothesis has some confirmation in some image analysis Paolo performed, which may show a tiny discrepancy in the arm's position after the last time we stowed it.
This would be relatively straightforward to solve -- not easy, but not dreadfully bad. We'd have to figure out where the arm actually is, then manually guide it through the unstow routine, at which point we'd be able to recalibrate the IDD by driving each of the joints against its hardstop. We might also need to repeat the "Martian Tai Chi" routine that recalibrates the camera models, so that the IDD and the cameras would be in agreement. It would be a long, slow process -- maybe a week or two, all put together -- but in the end we'd be as good as new.
But Eric's explanation isn't the only one that fits the observed facts. Ashitey fears there's something worse going on. One of the actuators in the testbed rover has started to fail, and it stalls out in pretty much the same way as we've just observed Opportunity's joint motor stall. If this is what's going on, it means the IDD has become unreliable -- every once in a while, our IDD work might randomly fail in this way. That would introduce a severe, somewhat random penalty into all IDD work.
With the testbed rover's IDD, the failures are merely annoying; repeat the failed command, and it usually works the second try. If Opportunity's IDD is indeed failing in this way, that means we have an occasional one-sol penalty to look forward to; probably nothing worse than that (though that would be unpleasant). But it could be worse still: the actuator might simply have failed permanently. And if that's what happened, a large fraction of Opportunity's scientific potential just disappeared forever.
There are other possible explanations, too, with varying degrees of benignity. Erickson wants the full-up anomaly investigation, and the right guy to do that is Ashitey. But Ashitey's RP-1 on Spirit today, which would leave him precious little time for something like that. So he and I switch places -- he heads up the investigation, and I go back to Spirit.
Which means I'm driving today -- driving to Miami. (I try to remember not to leave the left turn signal on.) This drive is reasonably simple, snaking around a few rocks before heading more or less straight toward Miami. Well, not quite Miami -- Coral Gables. See, I didn't have an exact target for Miami, just picked a spot in the neighborhood, and Coral Gables is a Miami suburb ....
Maybe you had to be there.
If all goes well, we'll cover a third to a half of the distance to Miami in this drive. (Maybe we'll stop for the night at a HoJo's. Sorry, I can't help it.)
I'd seen a birthday card in the Spirit sequencing room last week, but it hadn't registered on me until today. I pick it up and read it. I figured it was for someone on the team. Turns out I was right; it's for the most important member of our team. It's a birthday card for a one-year-old -- our one-year-old, Spirit. (Or maybe that should be a "birthsol" card, but I'm not picky.) From the nice (and clever) folks at unmannedspaceflight.com.
I love it. We should make them Spirit's godparents or something.
[Next post: sol 680, December 1.]
2010-11-27
Spirit Sol 676
Thisol should be easy. Spirit's been IDDing a patch of outcrop named "Seminole," on the target "Osceola." (It's coming up on Thanksgiving, and since we did Pilgrim names last year, we're doing American Indian names this year.) And the big science waypoint we've been heading toward has been renamed from "Waypoint" to "Miami." I'm getting a kick out of this, having gone to Seminole High School and all. But I digress.
So all we're doing is repeating the same stuff they did yesterday on Osceola, only we're changing the target to Abiaka (another Seminole chief). Load up those sequences, change the targets and a few sequence IDs, and you're done.
As is so often the case, it's not that simple. Abiaka unfortunately turns out to be a little too vertical (strictly speaking, Abiaka is a too-vertical normal to a too-horizontal rock surface) -- a problem because the IDD gets confused when you ask it to move straight up and down. That has to do with the target being too close to a wrist-flip condition, where the IDD could choose almost equally well between two different wrist configurations, and if conditions on Mars vary only slightly from the predicts, it could change its mind in the middle and decide it can't do what we want.
The remedy for this is known but tedious: tweak the angle of the target, so that it's tilted slightly away from vertical, until the simulation says you've gone far enough. What complicates this is that we can't tilt the normal too far, or the RAT brushing won't work. Also making it more complex is that John's having his shadow, Ashley, handle the IDD sequence, and while she generally knows what she's doing, she's still new to this and somewhat slow at it.
Oy. Well, it takes forever, but it gets done, and that's what matters.
[Next post: sol 678, November 29.]
So all we're doing is repeating the same stuff they did yesterday on Osceola, only we're changing the target to Abiaka (another Seminole chief). Load up those sequences, change the targets and a few sequence IDs, and you're done.
As is so often the case, it's not that simple. Abiaka unfortunately turns out to be a little too vertical (strictly speaking, Abiaka is a too-vertical normal to a too-horizontal rock surface) -- a problem because the IDD gets confused when you ask it to move straight up and down. That has to do with the target being too close to a wrist-flip condition, where the IDD could choose almost equally well between two different wrist configurations, and if conditions on Mars vary only slightly from the predicts, it could change its mind in the middle and decide it can't do what we want.
The remedy for this is known but tedious: tweak the angle of the target, so that it's tilted slightly away from vertical, until the simulation says you've gone far enough. What complicates this is that we can't tilt the normal too far, or the RAT brushing won't work. Also making it more complex is that John's having his shadow, Ashley, handle the IDD sequence, and while she generally knows what she's doing, she's still new to this and somewhat slow at it.
Oy. Well, it takes forever, but it gets done, and that's what matters.
[Next post: sol 678, November 29.]
2010-11-19
Opportunity Sol 649 (Spirit Sol 668)
Meanwhile, on the other side of the planet ... Opportunity is still skirting the rim of Erebus Crater, making her way to the potential entry point we can see from orbital data. Thisol's plan is to finish up a little IDD work, then drive about 50m -- mostly along outcrop -- toward a flat patch of outcrop we can see in the distance.
This plan is complicated by the fact that the flat patch of outcrop isn't so flat. Looking at it in stereo, we realize that it's actually a somewhat steep face, rising about 50cm over less than 2m. (The fact that the outcrop seems to have some delta-Z makes me think the science team might be interested in it, as it might provide some layered rock to examine. Cooper, ever eager to get moving, is comically reluctant to point this out to them, but I do it anyway.) The ends of the patch are still relatively flat, though, so we decide to aim toward those instead.
Our drive sequence is further complicated by the fact that we're heading at about 180, but need to turn for comm to about 115 -- a large counterclockwise turn, problematic on this vehicle because of the stuck right-front steering actuator. If we knew in advance what our heading would be for the comm turn, that would make it a lot easier to sequence. But we don't, because toward the end there's a slight change in direction: we'll be headed either about 175 or 185 when the time-of-day limit hits. And the only tool we have to make the sequence conditional on the vehicle's yaw gives awkward results around 180, because it gives its answers in the range of -180 to +180 (actually -pi to +pi, since it's in radians) -- nothing we can't sequence around, it's just one more damn thing.
All this still wouldn't be too bad, if Cooper or I were driving. But we decided to let Paolo do it, since he needs the experience. He actually handles it really well, making good decisions and so on. The only problem is that he's still somewhat new, and therefore somewhat slow, and this is a complicated sol, and, well ... it takes a long time. I sit with him pretty much the whole time and strive to stay cheerful and positive. Add that to the list of challenges.
Oh, but we're not through yet. Because I'd been sitting with Paolo, I hadn't had a chance to closely examine Brian's IDD sequence. During the CAM, I discover that the MI poker gets awfully close to a rock during a lateral move. There's less than 1cm of clearance in the prediction, which is within our error budget. Luckily for me, Ken Herkenhoff -- the MI PI, basically the designer and builder and owner of the instrument -- is sitting in the room. (He's normally one of the remote scientists, but is out here to participate in training the new crop.) So I put the CAM on hold, show Ken the situation, and ask his advice.
Now, the MI poker has a bit of spring steel near its base, allowing it to flex safely in just such circumstances as these. And the simulation does show that we have clearance, albeit not much clearance. What's more, we're first touching the surface near the potential contact point, so any error about the location of that point will likely be accounted for -- if our imagery is misleading about that, it should be misleading in the same way about the nearby spot we'll touch, and the sequencing is relative to the touched location, so the error will effectively magically disappear. All this taken into account, Ken's comfortable with proceeding as planned, so we go ahead. (What occurs to me later, as well, is that the simulation we're looking at is actually relative to the 1cm overdrive position -- which aleady accounts for our uncertainty about the terrain -- which means we'll likely have 1cm more clearance than shown. Even better.)
We also realize another place where there's a potential problem. We're in that phase where the Earth-Mars time difference requires us to plan several sols ahead, so we don't yet know the true state of the IDD at the beginning of this sol's sequences -- for the very good reason that that stuff simply hasn't happened yet. And since we're starting (we hope) with the MB in contact with the CCT, we have to disable self-collision checking at the beginning of the sequence in order to allow ourselves to retract the IDD into free space. The problem is, if the planned IDD sequences don't go as predicted, the IDD could start off sol 649 someplace unexpected, and disabling self-collision checking is not what you want to do under those circumstances.
Our solution is to remove the IDD sequence from the bundle -- we'll call in Sunday morning, after we've looked at the data, and we'll send this IDD sequence to the spacecraft only after we know it's done what we expected it to do. This potentially means we'll have to drop the IDD sequence as well as the subsequent drive, since the rover won't drive when the IDD is unstowed -- and it also means we've all got some work to do Saturday night and Sunday morning -- but that's a lot better than risking the IDD.
At long last, we complete the CAM -- and not a moment too soon, as it's been a long and grueling day and we're all getting kind of punchy -- and I realize we've made yet another mistake.
"We're not going to change anything," I say carefully, "but we did screw something up." Normally, our drive sequences detect absent helper sequences, precluding further driving if a helper is missing. (This would mean, for instance, that we were unable to perform a slip check; without that, we don't know whether the vehicle is bogged down, so we don't want to keep trying to drive for tens of meters, making the situation worse.) Our predicted downlink is so poor thisol, however, that we wanted to be sure we'd turn for comm even if one of the helper sequences didn't make it on board. That's rare, but it happens, whether due to a human screw-up somewhere in the process, equipment problems at the DSN stations, or simply because of interference in the long journey of the radio waves from Earth to Mars.
So we pulled a clever trick: during the sequence, we abuse the OK-to-IDD flag -- if a helper is absent and we preclude driving for that reason, we also preclude IDDing. Later, the sequence checks whether we're OK to IDD, and if not, that means a helper must have been missing, so we re-enable driving (and IDDing) so that we can turn for comm.
"What we forgot," I conclude, "is that if you try to drive while driving is precluded -- which we would -- you also get a goal error. And we don't clear that goal error, so the turn for comm still wouldn't happen."
But this bites us only if a helper doesn't make it onto the vehicle -- a rare event -- and it doesn't truly hurt us, it just fails to help us. And at this point on this sol, nobody wants to go back and fix it. So we don't.
[Next post: sol 676, November 27.]
This plan is complicated by the fact that the flat patch of outcrop isn't so flat. Looking at it in stereo, we realize that it's actually a somewhat steep face, rising about 50cm over less than 2m. (The fact that the outcrop seems to have some delta-Z makes me think the science team might be interested in it, as it might provide some layered rock to examine. Cooper, ever eager to get moving, is comically reluctant to point this out to them, but I do it anyway.) The ends of the patch are still relatively flat, though, so we decide to aim toward those instead.
Our drive sequence is further complicated by the fact that we're heading at about 180, but need to turn for comm to about 115 -- a large counterclockwise turn, problematic on this vehicle because of the stuck right-front steering actuator. If we knew in advance what our heading would be for the comm turn, that would make it a lot easier to sequence. But we don't, because toward the end there's a slight change in direction: we'll be headed either about 175 or 185 when the time-of-day limit hits. And the only tool we have to make the sequence conditional on the vehicle's yaw gives awkward results around 180, because it gives its answers in the range of -180 to +180 (actually -pi to +pi, since it's in radians) -- nothing we can't sequence around, it's just one more damn thing.
All this still wouldn't be too bad, if Cooper or I were driving. But we decided to let Paolo do it, since he needs the experience. He actually handles it really well, making good decisions and so on. The only problem is that he's still somewhat new, and therefore somewhat slow, and this is a complicated sol, and, well ... it takes a long time. I sit with him pretty much the whole time and strive to stay cheerful and positive. Add that to the list of challenges.
Oh, but we're not through yet. Because I'd been sitting with Paolo, I hadn't had a chance to closely examine Brian's IDD sequence. During the CAM, I discover that the MI poker gets awfully close to a rock during a lateral move. There's less than 1cm of clearance in the prediction, which is within our error budget. Luckily for me, Ken Herkenhoff -- the MI PI, basically the designer and builder and owner of the instrument -- is sitting in the room. (He's normally one of the remote scientists, but is out here to participate in training the new crop.) So I put the CAM on hold, show Ken the situation, and ask his advice.
Now, the MI poker has a bit of spring steel near its base, allowing it to flex safely in just such circumstances as these. And the simulation does show that we have clearance, albeit not much clearance. What's more, we're first touching the surface near the potential contact point, so any error about the location of that point will likely be accounted for -- if our imagery is misleading about that, it should be misleading in the same way about the nearby spot we'll touch, and the sequencing is relative to the touched location, so the error will effectively magically disappear. All this taken into account, Ken's comfortable with proceeding as planned, so we go ahead. (What occurs to me later, as well, is that the simulation we're looking at is actually relative to the 1cm overdrive position -- which aleady accounts for our uncertainty about the terrain -- which means we'll likely have 1cm more clearance than shown. Even better.)
We also realize another place where there's a potential problem. We're in that phase where the Earth-Mars time difference requires us to plan several sols ahead, so we don't yet know the true state of the IDD at the beginning of this sol's sequences -- for the very good reason that that stuff simply hasn't happened yet. And since we're starting (we hope) with the MB in contact with the CCT, we have to disable self-collision checking at the beginning of the sequence in order to allow ourselves to retract the IDD into free space. The problem is, if the planned IDD sequences don't go as predicted, the IDD could start off sol 649 someplace unexpected, and disabling self-collision checking is not what you want to do under those circumstances.
Our solution is to remove the IDD sequence from the bundle -- we'll call in Sunday morning, after we've looked at the data, and we'll send this IDD sequence to the spacecraft only after we know it's done what we expected it to do. This potentially means we'll have to drop the IDD sequence as well as the subsequent drive, since the rover won't drive when the IDD is unstowed -- and it also means we've all got some work to do Saturday night and Sunday morning -- but that's a lot better than risking the IDD.
At long last, we complete the CAM -- and not a moment too soon, as it's been a long and grueling day and we're all getting kind of punchy -- and I realize we've made yet another mistake.
"We're not going to change anything," I say carefully, "but we did screw something up." Normally, our drive sequences detect absent helper sequences, precluding further driving if a helper is missing. (This would mean, for instance, that we were unable to perform a slip check; without that, we don't know whether the vehicle is bogged down, so we don't want to keep trying to drive for tens of meters, making the situation worse.) Our predicted downlink is so poor thisol, however, that we wanted to be sure we'd turn for comm even if one of the helper sequences didn't make it on board. That's rare, but it happens, whether due to a human screw-up somewhere in the process, equipment problems at the DSN stations, or simply because of interference in the long journey of the radio waves from Earth to Mars.
So we pulled a clever trick: during the sequence, we abuse the OK-to-IDD flag -- if a helper is absent and we preclude driving for that reason, we also preclude IDDing. Later, the sequence checks whether we're OK to IDD, and if not, that means a helper must have been missing, so we re-enable driving (and IDDing) so that we can turn for comm.
"What we forgot," I conclude, "is that if you try to drive while driving is precluded -- which we would -- you also get a goal error. And we don't clear that goal error, so the turn for comm still wouldn't happen."
But this bites us only if a helper doesn't make it onto the vehicle -- a rare event -- and it doesn't truly hurt us, it just fails to help us. And at this point on this sol, nobody wants to go back and fix it. So we don't.
[Next post: sol 676, November 27.]
2010-11-17
Spirit Sol 666
Sol 666: the sol of the Beast. It's a real shame this didn't happen to be Halloween. Maybe thisol should be Martian Halloween.
If there's a Beast involved, he's being awfully good to us. Yestersol's drive went just as expected, a 42m drive -- not much farther than the previous sol, but in a shorter time, thanks to (ahem) remembering to turn on step-skipping during the autonav segment.
Other things are looking up, as well. Our flash volume and data-product count are both still high, but dropping, like the late stages of a Flash Flu. We still need to exercise some restraint; SOWG chair Wendy Calvin proposes alternating atmospheric science sols and geological science sols as a way to reduce flash pressure. Driving is the thing we won't cut -- at least, not entirely, though we do end up shortening the duration once again thisol -- since we have a definite goal and need to make continual progress. The sun is tracking northward in the sky once more, and it isn't going to wait for our little rover, no matter how much she needs and deserves it to.
Yestersol's drive left us about 40m from a small ridge, over which we can just barely see the current science goal (a pile of jumbly black stuff creatively named "Waypoint"). It's 150m away -- we had been thinking it was somewhat closer -- and with a clear, blind-drive-safe 35m to the ridge, we could have made 80m of progress if given three hours to drive (autonav willing). As it is, they cut the drive back to include only the blind part, and maybe a few meters of autonav. We'll make only 40 to 45m.
There's a lot of difference between those two numbers. But if that's all we can afford, that's what we'll do.
[Next post: sol 668 (Opportunity sol 649), November 19.]
Courtesy NASA/JPL-Caltech. The creatively named "Waypoint."
If there's a Beast involved, he's being awfully good to us. Yestersol's drive went just as expected, a 42m drive -- not much farther than the previous sol, but in a shorter time, thanks to (ahem) remembering to turn on step-skipping during the autonav segment.
Other things are looking up, as well. Our flash volume and data-product count are both still high, but dropping, like the late stages of a Flash Flu. We still need to exercise some restraint; SOWG chair Wendy Calvin proposes alternating atmospheric science sols and geological science sols as a way to reduce flash pressure. Driving is the thing we won't cut -- at least, not entirely, though we do end up shortening the duration once again thisol -- since we have a definite goal and need to make continual progress. The sun is tracking northward in the sky once more, and it isn't going to wait for our little rover, no matter how much she needs and deserves it to.
Yestersol's drive left us about 40m from a small ridge, over which we can just barely see the current science goal (a pile of jumbly black stuff creatively named "Waypoint"). It's 150m away -- we had been thinking it was somewhat closer -- and with a clear, blind-drive-safe 35m to the ridge, we could have made 80m of progress if given three hours to drive (autonav willing). As it is, they cut the drive back to include only the blind part, and maybe a few meters of autonav. We'll make only 40 to 45m.
There's a lot of difference between those two numbers. But if that's all we can afford, that's what we'll do.
[Next post: sol 668 (Opportunity sol 649), November 19.]
Courtesy NASA/JPL-Caltech. The creatively named "Waypoint."
2010-11-16
Spirit Sol 665
Ah, heck. We meant to make 60m of progress, and came up with only 37m.
How did this happen? Let's see, autonav imaging takes roughly 3 minutes per 50cm step, which means we should expect to take 20 such steps per hour, or -- oops -- 10m per hour. And we spent two hours on autonav, as predicted, and thus made only about 20m of progress in that time.
Both Ashitey and I were remembering that autonav makes more like 35m per hour. But this is true only when we turn on a feature called "step-skipping." Instead of "image, step, image, step," it does "image, step, step, image, step, step," or even "image, step, step, step, image, step, step, step." Since the "image" part is the slow part, this makes a dramatic improvement in how fast we can drive.
But since this is less safe, since the rover moves farther before taking images of the world around it, it's not the default.
Duh.
Thisol we decide we'll turn on step-skipping.
To make matters worse, we had a poor downlink, and we have more data products on board than we'd like. (We have to worry not only about the total amount of data on the rover but also the number of separate data products.) SOWG chair John Grant proposes cutting the drive short as a way of reducing the generation of new data products until some old ones can be downlinked.
"Nah," I say, "we're not gonna do that."
"But if we don't, the rover will start autodeleting old science products," Emily Eelkema chimes in.
That's what I was waiting for. Since it's been on my mind, I say: "Forget those old MI images! They're history! Let it go! Embrace change!"
They cut the drive short.
Well, it won't be too bad. With step-skipping on, we'll make the same amount of distance as today (about 40m) in the shortened time -- still a decent amount to show for ourselves.
As long as we're changing things in the sequence ... a significant contributor to the drive time is the periodic slip-check sequence. Since it has to invoke visodom, and all the complex 3-D processing that goes along with it, this tends to chew up a lot of duration. What's more, since we're now fairly confident we can make it all the way downhill, most of the original rationale for the slip checks (ensuring we'd be able to make it back to the top of Husband Hill if we got into a bad spot along the way) has disappeared. So I ask Ashitey what he thinks about taking them out.
"Maybe after the Martian-year anniversary," he says. It's next week. "We want both rovers healthy for that."
This is wisdom. I let it go.
Late in the day, Justin Maki points out an article on the Aviation Week Web site. It's something Craig Covault wrote about his recent visit. I didn't realize it at the time, but he was still here for our big discussion about IDDing Hillary.
Emily seems a little worried about the article, but I'm not. "I'm glad he was here for that," I say. "I think it shows us at our best. We had a real concern about a genuinely difficult and novel problem, we took it seriously, we did the analysis we had to do, we proceeded with the right degree of caution -- and in the end, we got the science."
Covault's article is hampered somewhat by the fact that he was on the other end of a teleconference, listening to unfamiliar voices. As a result, some of the details get smudged. But taking that into account, I think he got the big things right: our dedication, our talent, our -- well, damn it, I'm not ashamed to say our brilliance.
Because we're damn good at this. Even if we do forget to turn on step-skipping once in a while.
How did this happen? Let's see, autonav imaging takes roughly 3 minutes per 50cm step, which means we should expect to take 20 such steps per hour, or -- oops -- 10m per hour. And we spent two hours on autonav, as predicted, and thus made only about 20m of progress in that time.
Both Ashitey and I were remembering that autonav makes more like 35m per hour. But this is true only when we turn on a feature called "step-skipping." Instead of "image, step, image, step," it does "image, step, step, image, step, step," or even "image, step, step, step, image, step, step, step." Since the "image" part is the slow part, this makes a dramatic improvement in how fast we can drive.
But since this is less safe, since the rover moves farther before taking images of the world around it, it's not the default.
Duh.
Thisol we decide we'll turn on step-skipping.
To make matters worse, we had a poor downlink, and we have more data products on board than we'd like. (We have to worry not only about the total amount of data on the rover but also the number of separate data products.) SOWG chair John Grant proposes cutting the drive short as a way of reducing the generation of new data products until some old ones can be downlinked.
"Nah," I say, "we're not gonna do that."
"But if we don't, the rover will start autodeleting old science products," Emily Eelkema chimes in.
That's what I was waiting for. Since it's been on my mind, I say: "Forget those old MI images! They're history! Let it go! Embrace change!"
They cut the drive short.
Well, it won't be too bad. With step-skipping on, we'll make the same amount of distance as today (about 40m) in the shortened time -- still a decent amount to show for ourselves.
As long as we're changing things in the sequence ... a significant contributor to the drive time is the periodic slip-check sequence. Since it has to invoke visodom, and all the complex 3-D processing that goes along with it, this tends to chew up a lot of duration. What's more, since we're now fairly confident we can make it all the way downhill, most of the original rationale for the slip checks (ensuring we'd be able to make it back to the top of Husband Hill if we got into a bad spot along the way) has disappeared. So I ask Ashitey what he thinks about taking them out.
"Maybe after the Martian-year anniversary," he says. It's next week. "We want both rovers healthy for that."
This is wisdom. I let it go.
Late in the day, Justin Maki points out an article on the Aviation Week Web site. It's something Craig Covault wrote about his recent visit. I didn't realize it at the time, but he was still here for our big discussion about IDDing Hillary.
Emily seems a little worried about the article, but I'm not. "I'm glad he was here for that," I say. "I think it shows us at our best. We had a real concern about a genuinely difficult and novel problem, we took it seriously, we did the analysis we had to do, we proceeded with the right degree of caution -- and in the end, we got the science."
Covault's article is hampered somewhat by the fact that he was on the other end of a teleconference, listening to unfamiliar voices. As a result, some of the details get smudged. But taking that into account, I think he got the big things right: our dedication, our talent, our -- well, damn it, I'm not ashamed to say our brilliance.
Because we're damn good at this. Even if we do forget to turn on step-skipping once in a while.
2010-11-15
Spirit Sol 664
Many writers and philosophers have spoken of the same idea: life is change, the Eternal Now, you can't step in the same river twice, etc. This concept is logically sound, and I should embrace it. But I have a really hard time with change, especially when the team's changing.
The team's changing.
In a good way; at least nobody I like is leaving. We're bringing in a raft of new scientists, and they're sitting there monitoring the SOWG meeting. It feels weird to be, increasingly, an old-timer on this project that I still think of as the Fresh New Thing. But more, I'm worried that the new scientists are harbingers of unhappy days ahead, days when some of the scientists I've known and liked will be leaving.
I'm sure I'll like the new people, too. But it won't be the same. The moving finger writes, and, having writ, moves on.
We're doing okay on our drive template. Unlike the primary mission, the scientists are actually trying to keep ahead of the line on the graph that shows how far the rover needs to progress each sol. We'll take a hit at Thanksgiving and maybe another at Christmas, but for now we're ahead of the line, even after turning around and going back to spend a week or so IDDing Larry's Bench. Today we're going to make the situation even better, by plowing ahead another sixty meters.
The gaggle of new scientists has brought back a familiar face, Steve Squyres, who's still kvelling about our recent 100m drive (okay, 94m) and looking forward to this one. But, he points out, "You can't slide anything past this uplink team -- even if you try to blow past something interesting, they'll make you turn around and go back!"
Maybe I should figure out a way to cow the new scientists now, before they can develop a troublesome attitude. Change can also be seen as an opportunity, of course.
The team's changing.
In a good way; at least nobody I like is leaving. We're bringing in a raft of new scientists, and they're sitting there monitoring the SOWG meeting. It feels weird to be, increasingly, an old-timer on this project that I still think of as the Fresh New Thing. But more, I'm worried that the new scientists are harbingers of unhappy days ahead, days when some of the scientists I've known and liked will be leaving.
I'm sure I'll like the new people, too. But it won't be the same. The moving finger writes, and, having writ, moves on.
We're doing okay on our drive template. Unlike the primary mission, the scientists are actually trying to keep ahead of the line on the graph that shows how far the rover needs to progress each sol. We'll take a hit at Thanksgiving and maybe another at Christmas, but for now we're ahead of the line, even after turning around and going back to spend a week or so IDDing Larry's Bench. Today we're going to make the situation even better, by plowing ahead another sixty meters.
The gaggle of new scientists has brought back a familiar face, Steve Squyres, who's still kvelling about our recent 100m drive (okay, 94m) and looking forward to this one. But, he points out, "You can't slide anything past this uplink team -- even if you try to blow past something interesting, they'll make you turn around and go back!"
Maybe I should figure out a way to cow the new scientists now, before they can develop a troublesome attitude. Change can also be seen as an opportunity, of course.
2010-11-13
Spirit Sol 662
This was supposed to be an easy sol -- just an IDD tool change, a sequence I could crank out in a couple of minutes. It turns out to be even easier than that -- no RP sequences at all.
I spend the morning working with Ashley to build a candidate drive sequence for Monday. We don't get many opportunities to let an RP-in-training do something like this when there's no time pressure, so it seems like the perfect time.
When we're all done, we've got a sequence that Ashitey will build on for our first drive next week. And I go off to start clearing away the crushing backlog of other work.
[Next post: sol 664, November 15.]
I spend the morning working with Ashley to build a candidate drive sequence for Monday. We don't get many opportunities to let an RP-in-training do something like this when there's no time pressure, so it seems like the perfect time.
When we're all done, we've got a sequence that Ashitey will build on for our first drive next week. And I go off to start clearing away the crushing backlog of other work.
[Next post: sol 664, November 15.]
2010-11-10
Spirit Sol 660
Today we're planning two sols, 660 and 661. Spirit finds herself poised before an enticing chunk of outcrop called "Larry's Bench," and the question on the science team's mind is whether we want to IDD the part we can reach, or reposition the rover to IDD another part, just to the right.
At first, this seems like a silly question to me. How can this bit of outcrop be so much different from a chunk of the same stuff, a few centimeters away, that it's worth spending even one sol to reach the other bit? (Indeed, it would end up being two sols; VEX needs at least one of our uplinks.)
The answer, it turns out, is dust. The stuff in front of us is pretty dusty, and there's no very good surface to RAT-brush it. But the competing section is already less dusty and has a broader and flatter surface that may be more amenable to brushing.
But maybe we can brush the part in front of us ... and so the discussion goes on, and on, and on. The SOWG meeting, which we normally try to finish in under an hour, goes to an hour and a half. And a good chunk of it is spent going in circles, at that. Alicia Falacarro -- now Alicia Vaughan -- can't stand it when the meetings go this way. She looks at me, rolls her eyes, and groans, "Definitely the J.V. team today."
But the answer in the end is that we're going to IDD, not drive, and the IDD work is reasonably straightforward. (Ed Guinness names the IDD target "Thrasher" -- "since that's what we did in this meeting.") Indeed, Ashitey does most of it. It's your standard unstow, MB touch, MI-2x2x5+stereo, MB re-touch, RAT-brush, post-brush MI-2x2x5+stereo, MB re-touch, APXS kind of sol. Oh, and then another change-tool back to MB on the following sol.
Matt Heverly is shadowing me today, so I take the time to walk him carefully through the whole sequence and talk over his questions in detail. The price of this attention, of course, is that I put him on the hook to do the walkthrough at the CAM. And a good job he does of it, too.
Sol 661. Sol six hundred and sixty-one. Unfathomable. It's almost a Martian year -- a Martian year being 669.7 sols. They're going to celebrate it in von Karman next week.
Six hundred. And sixty-one.
And counting.
[Next post: sol 662, November 13.]
Courtesy NASA/JPL-Caltech. Larry's Bench, right between our wheels.
At first, this seems like a silly question to me. How can this bit of outcrop be so much different from a chunk of the same stuff, a few centimeters away, that it's worth spending even one sol to reach the other bit? (Indeed, it would end up being two sols; VEX needs at least one of our uplinks.)
The answer, it turns out, is dust. The stuff in front of us is pretty dusty, and there's no very good surface to RAT-brush it. But the competing section is already less dusty and has a broader and flatter surface that may be more amenable to brushing.
But maybe we can brush the part in front of us ... and so the discussion goes on, and on, and on. The SOWG meeting, which we normally try to finish in under an hour, goes to an hour and a half. And a good chunk of it is spent going in circles, at that. Alicia Falacarro -- now Alicia Vaughan -- can't stand it when the meetings go this way. She looks at me, rolls her eyes, and groans, "Definitely the J.V. team today."
But the answer in the end is that we're going to IDD, not drive, and the IDD work is reasonably straightforward. (Ed Guinness names the IDD target "Thrasher" -- "since that's what we did in this meeting.") Indeed, Ashitey does most of it. It's your standard unstow, MB touch, MI-2x2x5+stereo, MB re-touch, RAT-brush, post-brush MI-2x2x5+stereo, MB re-touch, APXS kind of sol. Oh, and then another change-tool back to MB on the following sol.
Matt Heverly is shadowing me today, so I take the time to walk him carefully through the whole sequence and talk over his questions in detail. The price of this attention, of course, is that I put him on the hook to do the walkthrough at the CAM. And a good job he does of it, too.
Sol 661. Sol six hundred and sixty-one. Unfathomable. It's almost a Martian year -- a Martian year being 669.7 sols. They're going to celebrate it in von Karman next week.
Six hundred. And sixty-one.
And counting.
[Next post: sol 662, November 13.]
Courtesy NASA/JPL-Caltech. Larry's Bench, right between our wheels.
Subscribe to:
Posts (Atom)
Posts
