2009-11-23

Opportunity Sol 296 (Spirit Sol 317)

We're continuing the trek out of the crater, today with Jeff and me at the helm. Mostly it's a straightforward drive sol, following the now-usual pattern of driving up to a point and automatically continuing if we're where we're supposed to be.

Steve wants to squeeze in a science observation that would cost 10 megabits. The drive is the only place we can scavenge for it, but Steve makes it very clear that the drive itself is the top priority. "If you can't do it, 'no' is a fine answer," he says.

"Well, let's see," I muse. "We could maybe save four on the NAVCAM panorama wings, go down to two bits per pixel there."

Jeff nods. "And one from the penultimate front HAZCAM -- go down to half a bit per pixel, like the last one."

Halfway there already, but we've already picked the low-hanging fruit. "Can you get by with five megabits?" we ask Steve.

"Nah," he says, "thanks anyway, but if we're going to do this, we'd need the whole ten or nothing."

Jeff frowns, thinking about it. "Well ..." he begins.

"No!" says Art.

As we start to get back to it, I think of a way to scrounge another megabit. I turn around and slide over to the phone. "If we --"

"No!" Art snaps again.

Jeff and I shrug and turn back to the drive. A minute later he looks at me. "Well," he says, "if we reduce the motor/IMU data ...."

Art's head snaps up. "No!"

Now I know how a puppy feels.

[Next post: sol 327 (Opportunity sol 306), December 3.]

2009-11-20

Opportunity Sol 293 (Spirit Sol 314)

"This is an historic turning point in the Opportunity mission," declares Squyres. "Literally a turning point."

What he's referring to is the beginning of our egress from Endurance Crater. We spent weeks getting as close to Burns Cliff as we safely could, days taking pictures -- and now we'll spend weeks getting away again. But we're not just getting away from Burns Cliff, we're leaving the crater itself. Endurance Crater will soon be a thing of Opportunity's past. This is sure to make a lot of Opportunists happy, those who are sick of being in the damn crater already.

We're at the tip of a sort of narrow "safety peninsula," a rough triangle of driving-friendly rock surface pointed roughly at Burns Cliff. Uphill of us the slopes are too steep for driving, and downhill the sand is too soft. When you're between Scylla and Charybdis, you navigate carefully, and so it is with us. This first drive will be very modest: 3.6m backward, mostly in a very gentle uphill curve, then -- if the rover is sure it's right where we wanted it -- we turn a little bit downhill and drive another 3.6m.

Squyres is ready to make tracks. There won't be much more meaningful science until we egress, so the faster we can (safely) do that, the happier he and the rest of the science team will be. So, as he should, he pushes on whether we can safely extend this drive. I tell him I think it's a bad idea: due to the narrowness of this part of the path, we need to take baby steps. We've got maybe 50m between us and the bottom of our egress chute, and the conservative drive we've planned is already covering more of that in one sol than we'd planned. The original plan was to take about 18 sols to egress, averaging maybe 3m per sol; we'll double that on this drive.

He relents, but later he comes up with another idea. "Since you've got the rover testing its position at the midpoint of the drive, did you consider doing another extension in the same way?" Yeah, I tell him, I did consider it, but I didn't think of it until too late -- it would have torqued the plan too much.

"I wouldn't have liked it anyway," Ashitey says. This is in keeping with his reputation as a risk-averse RP.

"If we'd had John Wright here, we'd have done it," I quip.

It's too late to change the drive, and since Ashitey would object to our trying anyway, we leave it alone. 7.2m tomorrow, then -- if we're lucky. It'll be a good start.

Once we're out of the crater and back on the plains, we'll be covering a lot more distance than that. Ashitey points out something interesting: Cooper's eager to set a new single-sol distance record, but he's going on vacation for a month or so, right around the time we're planning to egress. Meaning the odds are that much better that one of the rest of us could set the record instead ....

[Next post: sol 317 (Opportunity sol 296), November 23.]

2009-11-19

Opportunity Sol 292 (Spirit Sol 313)

Opportunity's been sitting at the base of Burns Cliff, taking pictures, so they haven't needed a rover driver for a while. We were released yesterday, and we're released today as well. We'll come back tomorrow -- Friday -- to do the three-sol weekend plan, which has a drive at the end of it (on Sunday).[1]

But Ashitey and I stick around long enough to weigh in on the preliminary version of the three-sol plan. The drive plan normally includes a penultimate HAZCAM image -- an image taken just before the rover takes its last half-meter (or so) step. This is used to support IDD operations, since it gives us a picture of what's under the IDD at the final position, telling us whether it's safe to deploy. But since we won't be doing IDD work, I tell them to skip it. This saves a couple of megabits of downlink volume, which we can use to get science data instead, so I try to do this whenever I can.

But Ashitey wants the image. Why? His argument seems to be: because we always take it. This leads us into a huge philosophical discussion in which I try to convince Ashitey that engineering exists for the sake of science, and he tries to convince me that the engineering side has some kind of responsibility to use up everything it's managed to get the science team to relinquish. I think I win. Or at least it's a draw: Ashitey convinces me that in this particular case we do need the penultimate image in some form, to support some engineering analysis he's doing. But I bargain him down to a half-bit-per-pixel image, which gives us about half the savings I was originally shooting for.

Somehow this also turns into a discussion about how different rover drivers differ on risk-taking. This is something we agree on: Ashitey and Jeff Biesiadecki are the most risk-averse; John Wright and Chris Leger are the most risk-tolerant. The rest of us -- Frank, Cooper and I -- are in between. That's a little odd when I think of it. I'm normally a person who takes things to extremes, good or bad. To be in the middle of anything is new to me.

We don't have to produce a drive sequence today, but we know what we'll be asked for tomorrow. So we decide to build the drive sequence today, at least a rough cut. While we're doing this, John Callas tells us he's bringing the current astronaut class around for a tour, and could we arrange something cool to show them?

One of the great things about being a rover driver is that you never lack cool things to show. (This is very different from being a software developer, where the most cool things about your work often don't look like anything special to a lay audience.) We just hang out in the sixth-floor meeting room and put RSVP up on the big projectors, so they'll display our work in progress.

About an hour later, maybe 30 or 40 astronaut wanna-bes pile in and find seats at the big tables and in what I think of as the audience section. The astronauts-in-training are an interesting mix. Maybe a third to half of the total, and nearly all of the men, are the classic Neil Armstrong pattern: ex-military types with buzz cuts and bulky muscles. The women are less uniform. One or two look like ex-military, one or two others look like somebody's mom, most of the rest look like any women you'd see at the grocery store. Two are ridiculously hot, but one of them makes the mistake of talking. I don't think she'll be flying the Space Shuttle. I hope the other one makes it into space. No, to be honest, I actually hope she flunks out of the astronaut corps and decides she was so impressed by RSVP that she wants to work on MER.

I may be a Martian, but I'm only human.




Footnotes:

[1] During the sols that I was off shift, a huge controversy erupted over the wisdom and feasibility of actually driving to the base of Burns Cliff. The upshot was that we might be able to get there, but the intervening terrain created a risk that we wouldn't be able to get Opportunity safely out of Endurance Crater again if we did. The only upside was an improvement to the quality of the imaging we'd get of Burns Cliff, and as scientifically valuable as that was, Opportunity's safety ultimately trumped that consideration. So we pulled up as close as we could safely get and started imaging from there.

I recognize that that was the right decision, as borne out by the tremendous science Opportunity went on to do afterward. We got great imaging of Burns Cliff and then got a whole lot more science after egressing Endurance, continuing to this very day. Despite all that, it still bugs the hell out of me that we never quite made it to that goal.

2009-11-10

Spirit Sol 304

The trivial IDD sequence went well, so thisol we'll stow and go.

Stowing is easy. The hard part is going. It's hard because Spirit is so power-limited, and the solar insolation maps we have don't show much hope for us on the way to our next target, the imaginatively named Dark Rock. These maps show us a false-color overlay on navigation images, with the different colors reflecting how much solar energy we'll be able to collect at any given location. (This varies with the magnitude and direction of our tilt: tilting toward the sun is good, away is bad.)

In these images, red is bad, blue is good, green is in between. (Where "bad" means the rover doesn't get enough energy and could, in the worst case, die.) General policy is to lay out paths that are blue all along the way, just in case the drive faults out. Failing that, we choose paths that let us "hop" from one blue region (which we call "lily pads" even though they're not green) to another.

But in this case, we're climbing a generally south-facing slope, and there's not much blue to be seen. We're OK for the first few meters of the drive, but after that the landscape is soaked in red.

Well ... when all else fails, cheat. The different colors represent numbers on a scale from 0 to 1, where red is used for anything below 0.85. (The real floor on the insolation number is supposed to be around 0.83, but due to uncertainties in the way the map is computed, we add a small safety margin and use 0.85.) But Jeff and I find that we don't have to lower the low end very far -- only to about 0.78 or so -- before we start to see small patches of blue. It's not much, but it might be enough for us to hopscotch uphill.

If we're allowed to lower the threshold, that is. So I ask Kirk Fleming and Jake Matijevic about this, stressing that I'm OK with telling science we can't get there from here. Kirk and Jake hem and haw a little while, and come up with an absolute bare minimum insolation number of 0.75.

It's just barely enough, but it's enough. So Jeff and I get working on the first part of the overall drive to Dark Rock. Thisol we'll back uphill 4m, then -- using the same drive-extension trick I recently invented for Opportunity -- if we're on track, we'll turn and drive another 2m downhill, setting ourselves up nicely for the next sol's ~5m drive.

Dark Rock is only 20m away, as the Martian crow flies. With all the zig-zagging, and the short drive times imposed by the tiny amount of available solar power, it will take about a week to get there.

Back in my day, we'd laugh at 20 lousy meters.[1]

[Next post: sol 313 (Opportunity sol 292), November 19.]




[1] And now, of course -- Spirit being deeply embedded in Troy, and possibly unable to get out -- we'd be happy with 20 centimeters per sol. Oh, my poor little rover.

2009-11-09

Spirit Sol 303

I'm back in the Spirit World, and today's my first chance in a long time to work with Jeff Biesiadecki. Jeff's one of the other rover drivers, a smart and easy-going guy who wrote large chunks of the mobility flight software. He's also the guy who recommended me to Cooper in the first place (Jeff and I had worked together briefly in the Operations Engineering Laboratory, when I started at JPL), thus setting me on the path that eventually led to this, my dream job from childhood.

So I like working with Jeff.

It's an easy sol. Yestersol they did some IDD work and inadvertently closed the APXS doors just before placing it. It's easier to make this mistake than you might think. You close the doors simply by rotating the turret to a certain position. If you're trying to approach a certain target and the only way to get there is to rotate the turret to that point, well, then, the doors close. (Just to make the point: if this were software, they'd call that a bug. When it happens with hardware, everybody just shrugs.) Back when we worked together practically every sol, Bob Bonitz drilled into me the importance of watching out for this, and indeed it's part of my personal checklist.

So thisol we'll just reopen the doors and re-place the IDD. As it happens, the doors closed when the IDD was switching from the MB to the APXS, so it's not an inevitable consequence of measuring this target; we just need skip the usual preceding MB touch. Since the MB touch itself worked fine yestersol, we just need to get the recorded joint angles out of the downlink. After that, writing the sequence itself is something we could do with our eyes closed.

Apropos of nothing, we've known for some time that Opportunity started showing a sudden power boost recently. The reason is something of a mystery, but the likely cause is simply that Opportunity happened to be in the path of a dust devil, which blew some dust off its solar panels. (Badly power-deprived Spirit could use some of that dust devilly goodness, but no such luck.) The MI images Opportunity took of her magnets show that they've been cleaned off as well, which is consistent with the dust-devil explanation.

My favorite explanation involves a homeless Martian with a squeegee and a bottle of Windex. Mark Maimone points us to another possible explanation: a cartoon showing the solar panels with "WASH ME" finger-written in the dust.

2009-11-04

Opportunity Sol 277 (Spirit Sol 298)

When the rover receives the uplink, it sends us back a simple tone to acknowledge receipt. That's called the nominal beep. There's also an off-nominal beep, in case the rover thinks something went wrong with the uplink.

Today we didn't get either the nominal or the off-nominal beep from Opportunity. This is unusual, but it can, and probably does, mean nothing -- the DSN made a mistake in configuring the antenna, or something like that. But it can also mean the rover's dead, so it's cause for some worry.

As usual, this prompts me to think about ways I might have killed the rover. "Maybe we drove it off a cliff," I suggest to Andy.

"No, we beep before driving, so that can't be the reason."

He pauses. "Of course, maybe you drove over a cliff afterward," he adds helpfully.

Thanks, Andy.

Naturally, Opportunity is just fine. We never find out what happened with the beep, but we get the downlink right on schedule, so nobody cares too much. And what a downlink -- 122 Mbits, a huge amount of data, some as old as sol 207. This downlink even gets us over the "Herkenwall," a huge amount of MI imagery due to Ken Herkenhoff.

More important to me, yestersol's tricky drive seems to have gone well. Very well. Suspiciously well, come to think of it: according to the data, we're less than 20cm from our predicted position. Frank and I hold off on reporting the happy news until we can confirm it against the images, but it turns out we needn't have been so conservative; we really did do that well.

Even better, we got the full 8m out of the drive. After all that worry about the exact size of the post-first-segment target zone, Opportunity was only 10cm from its center -- basically, right on predict.

So it's easy to see we're driving again today. As we work on the sequence, election results are trickling in, and the telepresent scientists report them -- with diminishing enthusiasm -- as the night progresses. They sounded just like this during the American League Championship Series games the Red Sox lost. Looks like a tough year for Yankees fans and Democrats.

[Next post: sol 303, November 9.]

2009-11-02

Opportunity Sol 276 (Spirit Sol 297)

Ashitey's drive went just about perfectly over the weekend, leaving us just where we wanted to be: on a broad lane of exposed rocks, like a cobblestone highway leading us straight to Burns Cliff.

But our first stop is the Bermuda Triangle, or Triangle of Doom. This is a broad triangular (duh) expanse of soft sand -- seemingly the same kind of nasty stuff we were caught in for the last a couple of weeks -- whose apex crosses the highway ahead. We're not sure we can cross it, so our first step, over the next couple of days, will be to drive to it so that we can investigate it up close. Whether we can cross it or not, we'll be able to get better pictures of the cliff from there, and maybe decide whether we want to proceed or not.

The triangle is only 15m away, the kind of distance I'm used to thinking of as a single sol's driving -- and an easy one, at that. Of course, there's a catch: at the most recent traversability meeting, Jim Erickson said he didn't want us driving farther than about five meters per sol. The rationale for this is a good one. We don't have a lot of experience driving on this exact terrain, and the price for failure is very high: if we slip off of the rocks into the soft sand downhill, there's no guarantee that we can ever get uphill again, or indeed out of the crater. Even if we can, we'll squander a lot of precious time trying. So I don't disagree with the rule, but if we follow it strictly, it'll take us three sols to do a single sol's worth of driving.

I think I can get it down to two.

I propose my idea to Andy and Frank. "I think we can pick a spot five meters out where we can say, if the rover makes it here, we'd be willing to push a button to say to go another three meters on the same heading. So why not tell the rover that? Pick a point to serve as our prediction for the end of the nominal drive, and tell the rover that if it's within some tight radius of that point, go another three meters."[1]

That automatically cancels out the biggest objection to longer drives, namely, the a priori uncertainty about the rover's behavior in this terrain. If we don't make it to the predicted point, we simply don't do anything additional. But if we do, then we know our predictions are reasonably accurate, and it's reasonable to extend the drive in that case. And in the admittedly unlikely event that we can do this two sols in a row, we can make it to the triangle one sol sooner.

Frank and Andy agree readily. But that leaves us with a project manager to convince. Andy calls Jim and asks him to join us downstairs, and Frank starts pulling up the relevant imagery so we can make some good choices about drive distance and an appropriate radius for the circle. This means we're ready for Jim's questions, which is probably part of the reason we're able to sell him on the idea without a lot of trouble. (Well, "sold" is relative. His official response, delivered with his characteristic suspicious squint, is: "Okay, let's do something dangerous." But in Jim Erickson terms, that's sold.)

So that's our plan: five meters of nominal driving, with an optional three-meter extension at the rover's discretion. The radius of the circle varies through the day. When I we talk to Jim, we're thinking in terms of 25 or 30cm. But the more we look at the terrain, the wider the circle grows, reflecting our increasing confidence -- 50cm, 70cm, 75cm.

Then Squyres gets into the act. He's eager to get to the triangle and loves the idea of pushing the drive as aggressively as possible. He also recognizes that the larger the circle is, the more likely the rover will be within it and decide to drive farther. So he pushes for a bigger circle, and gets us to go back to the drawing board and bias the drive uphill so that we have a larger safe zone. The circle's eventual size: 90cm.

Part of what bounds the circle's size is whether we'll end up on rock. But there's another limiting factor. If we're expecting (say) 20% slip, then we don't want to choose a circle that would allow 50% slip, even if the rover would be safe anywhere within that larger circle. The whole point of the extension is that if the rover ends up anywhere in the chosen zone, it's okay to do another three meters like the first five. But if it slips far more than we expected, then even if it's currently safe, the added slip means we don't understand what the vehicle is doing, and so we don't want it to drive on without our stopping and taking a look at the situation. I think Steve wanted a larger safe zone (indeed, I know he did), but he understands our logic and acquiesces.

After the CAM, he apologizes -- graciously, though unnecessarily. "Hey, I'm sorry for pushing you guys on the drive stuff."

"No problem," I reassure him. "It's always good to have more people we can blame if anything goes wrong."




Footnotes:

[1] Another example of just inventing a technique on the fly and seeing it suddenly become standard practice. Now we do this kind of thing all the time. I'd forgotten that there was a time when it was novel, not one of those background "of course we do that" things.