Opportunity Sol 394 (Spirit Sol 414)

Cindy Oda's one of the nicest people I know, and she's going through a really rough time. Her mom's teeth are in really bad shape, and they're worried about her getting a bad infection. But they can't put her under to fix it because of some medication she's taking, and they're afraid to take her off it. Meanwhile, her husband, Jeff -- one of the other RPs -- has had to fly back to Illinois to deal with his own ailing dad, who has a heart infection or something like that.

So they're unhappy. Things like this make me almost feel bad about how well things are going in my life right at the moment.

Meanwhile, uh, things are going well in my life. (Sorry.) We're doing another uber-drive thisol. Actually, it's more of a mini-uber-drive -- no autonav. If we trip the suspension limits, we'll just have to stop.

I spent some time last night working out a prototype drive, which of course Khaled and Brian and I end up hacking up somewhat. Last night I didn't have time to analyze the past drives, so I wasn't really sure what limits to specify for this one. As the day wears on, Khaled and I keep thinking of something else we should check, look at the data, and interrupt Brian to tweak something else in the sequence. Tilt, bogie/differential limits at various points -- all of it comes up for review.

What we saw during the drive to Trieste was a tilt high-water mark of 11.5 degrees -- very close to the 12-degree limit that would have stopped the drive altogether. Both the tilt and suspension limits for that drive were higher than we've seen lately, which makes sense: we've been driving south, along the ripples, but the Trieste drive took us due west, straight across the ripples. So we were climbing up and down the ripples all the way. It's a wonder Opportunity didn't get seasick.

Because this drive took us somewhat off our intended southerly course, we're driving back to Vostok at a southeasterly angle. This is the worst of both worlds, as far as expected vehicle articulation is concerned. Not only will our tilt vary just as on the Trieste drive, as we climb up hill and over dale, but also our suspension will articulate more when we drive across the dunes at an angle than when we drove directly over them. This is unfortunate, as it means we may have to choose suspension limits that are hard to distinguish from what would happen if we encountered a real obstacle.

And that's a problem, because our drive-direction imaging is none too good. Usually we have good range data out to at least 80m or so, but thisol it peters out at 65m, and that's if you're being generous. Beyond that, we can't reliably determine a hazard from here, and if we can't tell the vehicle what one looks like in a way that distinguishes it from a ripple, we're in trouble.

In the end we decide, of course, to play it safe. We use more-generous-than-usual limits for the part we can see well -- that is, the part up to 65m -- and the usual paranoid limits beyond 65m. We'll probably make between 35m and 65m of progress, and then we'll have a data point that will be useful for planning the next drive. The next drive is going to be a weekend uber-drive series that should take us all the way to Vostok, or mighty close, so this will be very useful information to have.

Before Opportunity can start her drive to Vostok, though, she has to make it out of the crater. ("Crater," hah. Next to Eagle and Endurance, Naturaliste is a dimple.) Yesterday, when Frank and I were working ahead, I initially sequenced this egress using visual odometry. The strategy was a tried-and-true one: have Opportunity use VO to keep careful and highly accurate track of her progress, and stop when she's out.

Then we came up with the idea of using new R9.1 functionality[1] that can check for tilt. This seemed to let us express more nearly what we wanted -- we could have Opportunity drive until she's flat, assume that means she's out of the crater, and then proceed toward Vostok. We liked this better than the first approach, because we were worried the rover would yaw as she backed up, and end up on flat ground but not in the target zone. But if she only checked for tilt, we'd succeed no matter where she ended up, as long as it was flat. So I rewrote the whole thing to do it that way.

But today, when Khaled and I look at the data more closely, we see some problems with that idea. The area immediately behind Opportunity is flat, but the area just behind that has a largish ripple, one it would be easy for Opportunity to end up on. Given that she might be there, what tilt value do we choose to represent a flatness threshold? Too small, and the limit might be tripped because of the ripple, even though she had in fact egressed. Too large, and she might think she was out of the crater even when she wasn't. Neither of those outcomes would bring us joy.

Well, the reason we switched from a position-based approach to a tilt-based approach was that we were worried about yaw. And when we look at what happened on the way in, we see that Opportunity didn't yaw appreciably. Since we're backing straight up, leaving the same way we came in, she probably will do the same on the way out. In other words, yaw isn't likely to be a worry after all.

So I change it all back. Sigh.

Well, one thing I know for sure: if that's the worst thing that happens to me today -- and it is -- I'm damn lucky.

I'm damn lucky.

[Next post: sol 419 (Opportunity sol 398), March 8.]

---

Footnotes:

[1] That is, new functionality that had been added as part of the recent (R9.1) flight software upgrade.