I just read on SPACE.com that Oppy managed to get all six wheels stuck deep in a sand ripple.


From the picture
http://www.space.com/php/multimedia/imaged...t%3A+NASA%2FJPL
the sand looks very fine grained, maybe a tad powdery. Squyers is optimistic they'll get out,
but it seems they'll have to do some rocking back and forth with prayers that they don't get in deeper. The taller ripples and the different textured sand is curious; any ideas out there as to why it gets that way farther south from the landing site?

I don't mean to be a cynic, but perhaps Oppy won't die of her own natural causes; maybe Mars, the god of war, has found prey....

I guess you havn't followed the 6-7 pages on the subject here
http://www.unmannedspaceflight.com/index.p...findpost&p=9426

The "news" is almost 4 days old now. But Space.com isn't that quick

At least we have a dedicated thread for this now! It's a little ironic that the discussion has been going on in a thread named "Erebus/victoria Entry & Exploration, Would it be worth it?" when that question seems awfully premature given the current predicament!

I just wish I shared Steve Squyres optimism about Oppy, I'm worried sick! But I guess he knows best, here's hoping

James

I better continue here, since Erebus and Victoria are faaaar away. Maybe we re-open it next autumn.....I couldn't sleep well last night. Maybe it is not nescessary, but the thought of a stuck rover that is functioning well got to me. Ah, we'll see, it probably is going to work out fine. Steve says so.

I used the time to do some thinking (also DVD, pay attention): the only reason why this happened has to do with the density of the material. It is considerably less dense than rocky soil, so it must be significantly higher in salt content, which is supported by the lighter colour and absence of blueberries. The white-ish mottled appearance in orbital images of the Etched terrain MUST be caused by locally elevated saltcontent of the material, instead of morphology. We're talking about physically wheatered (very finely fragmented) evaporite here! Evaporite wheathers easier than silicates, iron oxides or other mineral constituents. And that's exactly why we did not see any elevated outcrops on the way (nor on the horizon). The only outcrops we've seen is inside craters and at some points at the same level as the surrounding. The darker topsoil is definately much stronger (more resistant to wheathering) than the underlayment (evaporite). SO, i don't believe this fluffy material is global dust. It is formed in-situ.

Why ? Because there's forms of impact remnants in the white stuff on an regional scale (for example, Erebus). Therefore, the whitish material must have been there for a long time. While in the process of wheatering and levelling of, deeper parts got filled up with the darker material, mixing to some extent with the light stuff.

SO, there's only dunes here, all the way to Victoria (and probably further down). The fact that more and more (epsom?) salts are here, makes me think Oppy's on the edge of a former shoreline of some kind (lake, sea.....stream ?).

Oppy, follow the dark lanes please. It's safer !

Steves latest update.......
from http://athena1.cornell.edu/news/mubss/

April 27, 2005

A note to all you Opportunity fans: Get used to the current scenery, because we're going to be here awhile.

We are very optimistic that we'll be able to get out of here, but we're really going to take our time doing it. The first rule in a situation like this is "do no harm", which means that you don't rush anything. We're going to take lots of pictures of all the terrain around the vehicle, to get a very complete picture of the situation. We're going to do lots of testing with the rovers that we have on the ground to simulate the situation on Mars. This testing will be aimed not just at finding a plan that will work, but at finding the very best plan that will work. We may try quite a few small maneuvers with Opportunity that aren't intended to do anything other then help us gather more information... perhaps followed by even more testing. All of this is going to take a lot of time. But this is a very precious vehicle up there, in excellent health, and there's no reason to rush anything. I'll try to provide updates as the process moves forward, but the main message for now is to be very, very patient. No apparent progress in the images doesn't mean anything other than that we're being very careful to do our jobs right.

Exactly -- the evaporite layer was created as a body of water evaporated (repeatedly -- since there are many, many layers of evaporite) and deposited the salts and minerals suspended within the water. Since it's been a long time since there were bodies of open water on Mars, the evaporite layer has been there for a long time.

But I don't necessarily agree that there are nothing but dunes as we get farther south into the etched terrain. If you look closely at the super-resolution MOC images, you can make out the duning pretty clearly, and as dunes always do, they are alined and controlled by wind patterns. The lightest-albedo structures within the etched terrain are very definitely not alined along with the other visible dunes -- they are more like small ridges and "peaks" sticking out above the plains. More importantly, they *interrupt* the regular procession of wind-formed duning.

I'm not saying that these high-albedo structures are nice, clean, unmodified outcrops of pristine evaporite. But I think they are evaporite structures in the process of being weathered away. In which case, they'll probably look like big piles of very light sand with actual rock sticking out at various angles (and weathered rather smooth). I think there *may* be some relatively uneroded evaporite exposed as a ground layer around these little ridges, which is what I'm most interested in seeing and getting analyses from...

-the other Doug

http://www.unmannedspaceflight.com/doug_im...o448_tracks.jpg 580kb

Doug

Great image, Doug.

What do we see here? The duricrust seems to support the weight of the Rover as usual in the top third of the image, but apparently the wheels break through the surface and start churning after that point. The material below the surface is powdery but cohesive. One thing I don't understand is why the surface material seems to "squoosh" out from under the wheels, as in the middel of the top frame of this combinatin, and seen better in some other images.

Strange stuff. But I think this change in characteristics is significant.

--Bill

A CRACKING image, as you would say. Terrific detail!
Bill, couldn't it be they ran the wheel in the reverse direction at the point you mentioned? Or the wheel stalled for a moment?

Something definitely happened at that point in time. As we examine and analyse (we'll have time to do that _ad_nauseam_) we'll figure out what is happening. I suspect that we've been seeing differences in the soil profile here; the soil in this part of Meridiani is very thin anyway and the "A" horizon is the ubiquitous blueberry/sand/weathering byproducts so we aren't sure of what is just beneath our wheels. I'd volunteer to do some shovel trenches, after extricating our intrepid hero...

--Bill

I'd wager that what we're seeing in Doug's image is, at the transition between the two sorts of materials, a clod being pulled up onto the wheel and then another, then another, until suddenly the spaces between the clods start to fill and the wheel becomes wholly covered (and rather smooth). The way that there's a cyclic pattern which drops off into chaos at the end is *not* good, for it implies that the wheels are going to perform quite poorly when the time comes to try to get out of the rover-trap.

It almost reminds me of a snow crust. When you walk on it you are supported untill you hit a thin spot then suddenly your stuck up to your waist in a snow bank.
maybe the rover needed snow shoes.

I sought a view of this "funny dune" that Oppy maybe took sols ago. The next sol back with useful pics that I found is sol 439. I've made a Pancam-pan from this sol therein one can find this dune. Additionally I've labeled it in a smaller image, cut from the pan - with the "exact" site of crime , where I think it is.

Panorama (720KB) http://www.greuti.ch/oppy/oppy_sol439_pancam.jpg

Cutting http://www.greuti.ch/oppy/the_funny_dune.jpg

You can compare it by the Navcam pics from sol 439 http://marsrovers.jpl.nasa.gov/gallery/all...unity_n439.html and by my Navcam-pan of sol 446 http://www.greuti.ch/oppy/oppy-sol446-navcam-k.jpg

I'm not sure if anyoone has pointed this out, but if you look at the most recent front and rear hazcam images, it looks as if the rover was crossing a ripple at a very low angle. When it has crossed ripples earlier it was at an angle nearly perpendicular to the crest. The crests of the ripples are somewhat sinuous. The rover started across this one at a low angle to the crest. At about the point where the rover was crossing the crest of the ripple, the crest curved towards the left. The effect was that all of the left wheels ended up on one side of the ripple and all of the right wheels were on the other side of the crest. All six wheels ppear to be in the deepest (softest?) part of the ripple.

Good find. The "chalk outline" of our victim is a nice touch.

One observation: Look at you panorama and just before the scene of the crime I see an Anatolia-type of lineation immediately before the funny dune. This may not be relevant, but I noticed this first off. Also, there seems to be some layering in the dune just left of center.

Many clues here, Inspector Cleuseau...

--Bill

Could the change in consistency be in some way related to the appearance of the "minicraters" that were recently seen along Opportunity's route? Could they, perhaps, represent holes punctured (for some reason) in the crust, leading to collapse into a less dense layer beneath?

I'm thinking that Doug's picture in reply #7 doesn't look like a change in consistency of soil. To me, it most definitly looks like one or more leading wheels locked up causing the trailing wheels to churn soil up and dig in.

It seems likely that a change in soil would give some indication in the adjecent, undisturbed soil.

If a wheel had locked up, the rover would have stopped driving immediately - and there would be engineering data to show it happening.

That it ploughed on for a few feet into the dune, and that there is no metion of such data - rules out a 'lock up' imho

Doug

Hopefully there are many significant clues to find. If we get the Rover refloat again, at the moment I dont know where could be the right (save) path for Oppy to drive along.

BTW: To the right of the funny dune there are unusual high and wide dunes - maybe also a clue for a unusual location

Anyone whos been sailing will know you cant sail straight into wind - you have to tack. If it appears you cant drive straight over a dune - we will have to do the same I guess

so - they can simply tax along the line of the dunes, then switch back

\
/
\
/
\
/
\
/

etc

Doug

I'm thinking the same, probably in future we should avoid driving over dunes.

Maybe we should ask a member of the current world power of sailing http://www.alinghi.com/en/

That's right. When the dunes started picking up many sols ago, I thought that Oppyt would have to start tacking instead of driving over the dunes, but she was making good progress and didn't need to. Come to think of it, this change in the driving surface may be caused by no more than changes in the particle size distribution from aeolian sorting on the fore and lee sides of the dune.

No Martians in black suits...

--Bill

I noticed some of today's pancam sun images look like a smeared time exposure. Has anyone noticed this before? Could they be using such a technique while trying to spin the wheels in order to see how much motion they get?

http://marsrovers.jpl.nasa.gov/gallery/all...DIP2620L8M1.JPG
http://marsrovers.jpl.nasa.gov/gallery/all...DIP2620R8M1.JPG
http://marsrovers.jpl.nasa.gov/gallery/all...DIP2600R8M1.JPG

I've seen that before. I don't know why its that shape, maybe someone who does astrophotography could explain it.

I'd like to second those who are moving toward the idea that Oppy's problem is not that it has gotten stuck in a suddenly-soft patch of material, but rather that a hardware and/or software and/or command malfuntion caused it to attempt to drive with one or more sets of wheels turned in the wrong direction and/or frozen. After all, we know that Oppy HAS had a recent problem with it's right front wheel, which may be a symptom of a deeper underlying problem (electrical power distribution?) that is now beginning to affect other wheels. At any rate, this would help explain why they are taking so long to get Oppy moving again. If she were just stuck in soft material, the obvious (and perhaps only viable) response is to align all 6 wheels (in so far as possible) and attempt to back out slowly.

Doug, you are correct in saying that such a lock up should have given notice by other means -- but only IF everything else is working properly, including the human operators. In other words, given that we are nealy a year and a half into a grueling mission, do not discount the possiblity of cascading software/hardware/operator errors having caused us to stuck for a bit.

Glenn

P.S. I am quite sympathetic to the idea expressed at the beginning of this thread that a lot of water has passed under the bridge elsewhere on this topic, and that it behooves us to poke around a bit before jumping in? Hope I have followed my own advice. But of course there is also the difficulty of knowing when a new thread is appropriate -- some type of cross referencing would be cool, or maybe such already exists.

The problem that oppy had was with a steering actuator - not a drive wheel.

If it had been a hardware/software error on one or more drive wheels - we'd have known about it by now - it would have come down in the very first comms pass.

Notice that BOTH sides of the vehicle have dug into the dune, and the front two and rear two wheels show evidence of having spun in situ (the tread is full of dust).

All the physical evidence shows that simply - it drove into a dune. End of story - there's evidence to suggest that it could NOT have been a hardware failure of one or more driving wheels. It would have had to have been both rear wheels ( the ones in the direction of travel) locking up to start diging the trench, but they cant have locked up or we wouldnt see their tread rendered smooth by spinning in the dust.

The reason they are taking so long to get moving again is very VERY simple. They dont want to make it worse - and to ensure that they dont - they're taking the time to do lots of remote obs and further analysis of mini-TES whilst running test-bed scenarios - arguably the only sensible progression.

Doug

On the topic of "cascading software errors", this can't happen. Software doesn't get old. Computer hardware does, but can not cause subtle changes in behaviour. This is clearly not software or computer hardware related. I design and program computer systems to control machines. There are levels of safeguards in a spacecraft that wouldn't allow a software glitch of the type being discussed.

On the topic of the problem being related to "electrical power distribution" or something similar, there are feedback to the computers that can never provide a false positive performance. The complex signals can not be mimicked by faulty hardware. If that feedback system were to fail, it would be detected and the rover would stop. Thus, none of the wheels were pointed improperly and all of the wheels were turning properly. In fact, the rover did cut it's drive short because it did detect a problem.

So the rover was doing what was expected. And Squyres is always forth coming with that type of information, anyway. The fact that it has not been stated means it didn't happen. Remember Spirit's brain freeze. They told us about and how serious it was. This is not that serious.

Being a change in the soil, it might or might not have happened. The rover could be operating properly and still not do the right thing. They are still learning how to drive with the stuck wheel. As I noted earlier, it appears that the rover may have been trying to make a left turn (with respect to the drive direction). Combine that with the stuck wheel and softer soil and maybe it was just enough to start digging in.

ed

Software doesn't get old, but programmers make mistakes. They've been tweaking the driving software for some time now, and one of the recent changes (to compensate for the drive actuator failure, I belive) did indeed cause the type of software glitch being discussed, causing the rover to veer off to the left for a few metres before the drive stopped prematurely.

Now I'm not saying that I know for sure that a software bug was to blame, but it certainly can't be ruled out yet.

Hey, bugs happen. That's why spirit had it's brain freeze. But that doesn't mean the rovers can drive on in the face of hardware failures. And the software does not accumulate errors due to use or age. If they made changes that caused the problem, that falls squarely into the "learning to drive with the stuck wheel" category. It appears to me from the images that there was a turn to the left. Not being able to turn that wheel would make such a turn rather disasterous in soft soil.

They will figure that out, make the changes, test them completely, and drive out. In the mean time, enjoy the view.

I'm almost sure they are overexposed sun images, with smear due to CCD sensor and not rover movements... However, I do not know why they overexposed pictures.

Yup, just CCD bloom I think.

Regarding the sun images that have CCD blooming.

The purpose of the images is to measure atmosphere opacity, and to precisely locate the sun in the sky <relative to rover pancam mast camera coordinate system angles>

The exposures are commanded, not "auto-exposed". If they calculate wrong, or if the atmosphere is clearer than expected, the camera saturates pixels in the bright areas and the electrical charge in the saturated areas overflow into the non-saturated adjacent pixels.

I believe that this particular CCD <correct me if I've got this mixed with another planetary mission camera> can actually still give pretty good atmosphere opacity data from the overexposed images since the electrical charge is pretty much all there, just spread out.

I've a little question about software while we wait...

How in-synch do they keep Oppy and Spirit's software? I know there have been pauses for code updates in both rovers, but in general, do the tweaks made for Spirit's driving challenges remain separate from Oppy's?

I don't know what their architecture is. I'd assume that the code base for both rovers is the same, but the driving parameters/decision variables for each are different - so the slip tweaks done for Spirit are made in a "data file", which is different for Oppy. But the code that makes the decision for shutting off on a slip of X% would be the same for both. I did a quick Google and didn't find anything specific on the guts of the software.

Hopefully they don't have to manage two (slightly) different code bases for each rover. Even with the cool code management tools available now, that kind of environment can let unexpected bugs slip in.

It is my understanding that the flight software on both rovers are identical. The differences are going to be in settings and instructions.

Such things as "disable steering on right, front wheel" is a setting. Both rovers have the setting. On one it is on and the other it is off. The types of automated drives and manuvers that are allowed would be instructions. They have extended the types and cababilities of drives through software updates. There have only been two software updates that I am aware of. They are lengthy to upload (two or three days) and a tad bit risky. So they don't do it very often.

The operating system for the rovers is VxWorks. The flight software is custom written for the rovers. The "buy-off" for a software update is not going to be trivial. The chances of a bug slipping through are small though not negligable. The brain freeze early in the spirit mission was caused by a bug in VxWorks. It was not detected until spirit was near death. The amazing thing is that the software was designed well enough that they were able to recover the vehical and fix the problem. Having some experience with remotely maintaining and debugging machine control problems, I can say that such a recovery is a tribute to the software design. Truly well done.

ed

I think we might be over thinking this. It could be simply that the riover got high centered and the new more aggressive software (for longer drives) caused it to dig itsself in. Can the wheels tell if there is slip if all 6 are slipping the same amount? Dosen't it take differences in travel to account for slip? Or maybe the rover dosen't account for slip right before it makes a turn. Was the rover making a turn when it got stuck? If it had not tried to make the turn would it be stuck right now?

scott

I have a similar question.

When djellison explained current wheel directions, he wrote:
"The wheels will be pointing in funny directions because at the end of a drive they turn the rover in place to put it in the best position for the UHF passes that afternoon and the following morning."

Does rover make turn in place at the end of drive automatically, even when it encounters an obstacle? Oppy went a half of planned day distance and sensed unexpected resistance. What instruction does she have for this issue? Is it "make turn in place for good communication"?

On the contrary, if they turn a rover by a command: did they send this command because they haven't got information about the problem yet, or did they decide to take close look at the problem after the turn?

Id love to see what's going on at JPL with the test rover. They must be building lots of sand dunes Hopefully they'll post some pics.

http://www.pasadenastarnews.com/Stories/0,...2848411,00.html

He said the earliest Opportunity would be instructed to try a maneuver that was successful in the test-bed would be Thursday for Friday

Does that mean they have tested a maneuver in the testbed that got the rover out of a dune?

Um....I think so. If they did not succeed in the testbed, this statement would be premature I guess.

Thanks for the explanations, Ed. They take much of the "black box mystique" out of the Rover software.

I wonder if one of the tweaks made to Oppy's obstacle avoidance algorithms for travel in the dunes could have made it "ignore" the dune it was on when it tried to make an end-of-drive turn in place. I imagine that the obstacle avoidance algorithm reads the hazcam/navcam images and detects shapes to avoid. A sea of dunes would present a sea of potential obstacles to avoid and the likely put less weight to avoid the dune-shapes.

She'll get out and on the road again...

--Bill

Yes, the fact that the rover was able to first figure out that its constant rebooting wasn't doing anything, and then know to start sending diagnostic info - essentially asking for help - shows some really good design. Except for that bug that let it screw up in the first place.

I cite from the JPL MER site: "We planned a drive of about 90 meters (295 feet). After driving about 40 meters (131 feet), Opportunity dug into soft dune material"

Ergo it can't be a regular end-of-drive turn, unless the software will try doing one even when the drive ends prematurely, if so I think this is more of a bug than a feature.

tty

From the above-referenced article:

"The rover stopped driving when its remote sensing capabilities noticed it had not been able to make sufficient progress on a planned turn while it was driving backward."

That would seem to suggest that it still attempted a turn to help align the antenna after it had prematurely terminated the drive.

I took the sentence less optimistically to mean that they didn't necessarily have an approach that worked (it does note that they've been working on getting the consistency of their surface material right). I thought it meant only that if they came up with a maneuver that worked in the test bed, Thursday or Friday would be the earliest that they could send the commands. On the other hand, I assume that they wouldn't rush off to try the first thing that works, either, so this may just mean that they won't be sufficiently satisfied with the option until later this week. Guess we'll find out when we're all of a sudden peering back at those skid marks.

I also thought Oppy had tried to do the normal turn at the end of a drive, irrespective of a full drive or prematurely cut short drive. You can bet your bottom dollar that they have a manouvere that will get Oppy out of the dune

No, it says that the drive was aborted when it noticed that the turn wasn't working, not that it aborted the drive then tried to turn. I think the planned turn mentioned was supposed to be mid drive.

It's interesting that the slipping during the drive was not noticed, it should have been shoulden't it? I'd suggest that they write some code that looks at the wheels in the hazcam images to watch for the leading ones disappearing under the dunes!

James

It definitely did not attempt a turn in place at the end of the aborted drive. When a driving anomily is encounted, all further manuovers are aborted. They were going for 90 meters, it aborted at 40 due to lack of progress, therefore it would not attempt the turn in place. That's no big loss - optimal positioning is not critical in the face of an anomily.

On the issue of why did it drive so far when it was in trouble...

They can not detect slippage in any of the wheels while they are driving. They command wheel rotations. The motors are these very powerful geared-down servos with encoders for mileage. These wheels will turn when they tell them to. Even if they are just spinning in place. Sometimes they draw more current than others, but they always go on command. Sometimes slippage can draw more current and sometimes less than normal. In addition, they can't take pics on the move. optical odometry is the only way they can detect slippage.

I think scott nailed it when he pointed to the aggressive driving software. They don't stop often enough for a look around. And these are the consiquences. That's not a complaint. It's just the risks of going fast and furious.

ed

they CAN detect wheel slippage. They have inertial navigation which can be tallied with the rotation of the wheels - how else could they give us such specific slippage figures and how else would it complete a drive despite slippage occuring - such an incidednt would leave it short of the target.

And actually - I think it probably just did an end of drive wheel-wiggy just to settle itself after aborting the drive.

There is no blame to be had or pointed anywhere in this - it was probably a blind drive at the time (and 80m drive would probably be 60m + 20m autonav I'd imagine) and I can see why - all the dunes look the same and while sure - the tracks have been a little deeper over some dunes - we've not seen anything like this in more than 2km of Meridiani cruising, and there was nothing to mark out this dune as being different to any of the other 500 or so we've probably driven straight over.

Doug

The problem arose in the last meter or 2. After 1000's of meters across these dunes without trouble, i can understand that they don't stop every meter to visualy check the status of the rover. It all went well, the physical appearance of the dunes did not chance, and suddently this powder comes up. It is bad luck. I wouldn't call it going fast and furious. Nobody could have seen this coming. But lessons have to be taken from this: if oppy gets out safely (which i think she will), we better be prepared, that Victoria is going to take a looong time to reach, because the driving from then on will have to be very conservative. From now on, they actually have to visualy check every move they make!

Any changes in the strategy for driving the rover will depend upon what has caused its current problem. Several questions come to mind:

What exactly caused the wheels to dig in? Did the front wheels dig in because the density of the surface material was so much less? Was the angle of attack and specific load on the wheel different in this approach than with other dunes that have been crossed? To what extent have conditions changed or was this the result of a bad combination of previous condition? Is this situation the result of unique conditions that can be avoided?

If soil deposits are building up on the rover’s approach side of the dunes only we may be able to adjust the angle of approach to float lead wheels over the softer areas with less load on the approach wheel by varying the approach angle. Assuming that the windward side of the dunes have less low density topping, it may be possible to move parallel with the waves with greater confidence. Crossing over dunes would require more guidance from home.

Do we have adequate mapping of the etched terrain to find a track through the maze that leads to the crater?