People rightly marvel at modern surgical techniques that let surgeons leverage the power of robotics to repair the smallest structures in the human body through wounds that can be closed with a couple of stitches. Such techniques can even be applied remotely, linking surgeon and robot through a telesurgery link. It can be risky, but it’s often a patient’s only option.
NASA has arrived at a similar inflection point, except that their patient is the Mars InSight lander, and the surgical suite is currently about 58 million kilometers away. The lander’s self-digging “mole” probe needs a little help getting started, so they’re planning a high-stakes rescue attempt that would make the most seasoned telesurgeon blanch: they want to use the lander’s robotic arm to press down on the mole to help it get back on track.
All About Friction
We’ve been covering InSight for a while now, starting with a discussion of how the mole, part of the Heat Flow and Physical Properties Package (HP³) experiment, is supposed to work. As a quick recap, the mole is basically an electric impact driver, with a rotating cam that loads a spring to release a burst of mechanical energy into a heavy hammer. The impact is supposed to drive the mole into the Martian regolith a few millimeters at a time, slowly burrowing up to five meters into the regolith while dragging a sensor-filled tail behind it, to measure subsurface heat flow.
Unfortunately, HP³ has been unable to dig itself into the soil. The failure has been attributed to everything from hitting a rock just below the surface to a previously unknown layer of duricrust, a hard layer made of soil particles that have been cemented together by chemical precipitates in the water that once flowed freely on Mars. Add to these possibilities the fact that the Martian soil has proven to be far less cohesive than originally thought, giving the mole little friction to work with, and it’s no wonder it’s stuck.
Back in October of 2019, it looked like the mole was on the move again. JPL engineers had decided to use the lander’s robotic arm to press against the hull of the mole while it tried to dig. The thought was that increasing the friction would give it the boost it needed to penetrate the duricrust and get on the move again.
Whack-A-Mole
Initial reports were that the mole was making progress after this pinning maneuver, but that proved to be optimistic. The mole did make progress, but it popped back out of the hole on two occasions. Those failures cleared the way for the current and riskiest idea: using the robotic arm’s scoop to push directly on the back cap of the mole.
On the face of it, this seems to be the approach that makes the most sense. After all, if the mole is having trouble moving downward, applying force downward should help it penetrate the duricrust. But it’s not as simple as giving the mole a little shove.
First of all, consider the physical aspects of the problem. The mole is about 3.5 cm in diameter, which is a pretty small target to hit with the robot arm’s scoop. The back cap is also not the most friendly surface in terms of manipulation, either. It’s bisected by the mole’s tail, which looks like a flexible Kapton PCB. There’s also a protrusion on the back cap to one side of the tail, which would appear to limit the scoop’s ability to bear down on the mole.
The tail itself is a problem as well. Extreme care will need to be taken to ensure that the scoop doesn’t touch the flexible PCB, which would likely be damaged if it got pinched. The tail not only powers the mole but provides data connections; lose any of those and you’ve likely lost the mission. Positioning the scoop will have to be done very gingerly, and is already being practiced on Earth-side mockups and using the actual hardware on Mars.
Latency is certainly going to be a huge problem as well. The current round-trip time for radio signals from Earth to Mars is in excess of six minutes, meaning that realtime control of the operation will be out of the question. The moves will have to be carefully planned in advance and made in very small increments, so as to minimize the chance of damage.
Snatching victory from the jaws of defeat through engineering ingenuity has been the story of space exploration since its very earliest days, so there’s a good chance that engineers will be able to push on the mole without damaging it. Whether that game of interplanetary Whack-A-Mole will yield results is anyone’s guess, but even if it HP³ never gets a chance to dig down into Mars, it won’t be for a lack of trying.
I am certainly not an aerospace engineer, but am, nonetheless, dumbfounded that the design used kapton as a cable material. And kapton sheets do not have four degrees of flexibility that a wired cable has. So what am I missing?
They probably use that same arm to place it and get in started, my guess.
I suspect thin flex tape in place of copper cables was a weight and space constraint choice – flex stuff can be rolled very small and is pretty darn light per unit length. Its also probably a PCB in its own right all the way down – sensors in the tail imply some chips etc in it. Much more durable to have more consistent flex along the whole length than copper cables-solder/connectors-sensor pcb-connectors-cables repeat as needed which will have significant stress focal points at every join.
FFCs (flexible flat cables) are more reliable (bend many more times before failure), have less mass, can carry a large number of conductors in a very small space and are generally more durable. It’s no surprise that NASA went with FFCs.
It’s weight. The full 5 meter tether is only 75 grams, including integrated temperature sensors and markings for determining how much has been spooled out. There are 14 sensors integrated on the cable itself, and they’re resistive sensors in a 4-wire configuration. That would be a *ton* of wires.
You might also imagine that it’s size, as you can coil a flat cable tighter than a rounded one, but in fact the tether wasn’t tightly coiled at all. The full design document for the HP3 package is available here.
There is a point where shaving off material to reduce weight causes more problems than it solves.
Witness the “paper-thin” wheels (what were they thinking!) on the Curiosity rover, where sharp rocks have torn holes through.
Apparently Perseverance has redesigned wheels.
Whilst on the topic of rover wheels, ESA’s ExoMars’ skinny wheels strike me as strangely (ok ridiculously) low on ground contact area. Surely they were aware of Spirit’s trials and tribulations being bogged in martian sand traps?
Did the cable cause problems here? No. It wasn’t the problem. It’s making an attempt at a jerry-rigged solution to a problem they didn’t plan for harder. But that *always* happens whenever you’ve got a situation like this, where you just have to work with what you have. I’ve done wacko-crazy things that I never would’ve anticipated with isolated electronics to try to get them to work.
A flat flexible cable is obviously the right solution for this problem. Individual wires have their own problems, many of which the people on here have probably never thought of. Normal PVC insulation (used in most standard wires) falls apart under these temperatures and UV loads. PTFE insulation cold-flows if it’s under tension, so no spooling it up tightly. (On spacecraft you can’t even *tie* wires together tightly – the tension will slowly eat through the insulation and cause it to short).
Doing boreholes in space is Hard. The track record is really bad, although multiple times it’s just bad luck (Beagle II, the Philae lander, both of which ‘crashed’ in one way or another). Even with humans drilling the holes the cable tether’s broken once (in one of the Apollo missions).
“Witness the “paper-thin” wheels (what were they thinking!) on the Curiosity rover”
The wheels weren’t thin due to weight concerns. The difficulty of driving on Mars is that it’s like a sandy beach with hidden jagged (huge) rocks, which is a really difficult combination to deal with. It’s *not* an easy engineering challenge. The thin skin is so the tire can flex.
All the Mars rovers are basically built to deal with “average Mars”, which means that if you encounter an area that’s more sandy than normal, you’ve got problems. If you encounter an area that’s rockier than normal, you’ve got problems. Same deal. In Curiosity’s case they actually encountered rocks that were basically icebergs: jagged spikes with huge rocks underneath. Plus the area they’ve been driving over is far less sandy than they expected, so the flexing happens more often, and metal fatigue is wearing the wheels.
Remember Opportunity and Spirit got stuck in sandy soil for a while. Curiosity probably would *prefer* that kind of soil, but they didn’t get it.
This mechanical engineer is shocked too. No mechanical sheath for the kapton, unless I am missing something. What could go wrong?
I don’t understand why the engineers at NASA never thought that bracing the thing to get it started would be a good idea.
You don’t think they just dropped the little guy out like a butt nugget, do you? Surely it wasn’t dangled by the flex cable until it dug in.
***butt nugget***
Sorry – an egg.
I actually preferred “butt nugget”…i just wanted to highlight the term because I had nothing constructive to add.
Who says they didn’t? Limited operational budgets, weight and understanding of Mar’s surface mean this might have been the ‘best’ solution for what was known at the time. It might even work exactly as they expected a few KM away in different dirt, with how little we know about the geology of Mars nobody can say yet.
Hopefully this bodged job works, but if it does wreck the digger it hardly matters – it wasn’t going to produce results anyway, just as long as it doesn’t damage the rest of the rovers science potential its still a win.
Because it was pushed into the soil in the first place. The only reason why they need to brace it now is basically because they kept messing with it (because it wasn’t digging any farther).
Sort of. The penetrator started out in a “silo” or sorts, placed on the Martian surface by the arm. The silo was to keep the penetrator upright for the first 15 or so centimeters of its trip down. The whole housing was supposed to stay in place over the borehole. The only reason we’re seeing the top of the mole now is because they picked up the housing and moved it when the mole got stuck. You can even see the marks from the housing’s footpads in the Martian dust.
I’ve seen comments from the NASA operators that there were guiding springs in the support assembly, and I’ve never seen a cutaway of the support assembly to figure out how those were oriented. I know when they were talking about removing the assembly they were worried that if it was still in contact with the back of the mole (providing tension), moving the assembly would cause the Mole to back out more (they were also worried it might be caught on the assembly, too).
My guess was that the support assembly basically had springs that absorbed the initial recoil to allow it to dig into the ground (pushing on it), but again, without seeing a cutaway of the assembly, not positive.
Train some machine learning code with machine vision for execution on the unit, train it on the earth-based lander and then run it on the landers processor?
Why is machine learning/AI the default answer for almost anything these days?
A 6 minute round trip is slow but not prohibitive. Even with an abundance of caution, it shouldn’t take THAT long for engineers to position the scoop, almost certainly less time than it would take to develop a ML/AI model based on hardware that was never intended to run such programming (cameras/sensors not in an optimal location for the task), verify/validate/train and send such code and then STILL have the chance that it’ll damage something in the process.
Would you want to be that engineer with his hands on the joystick with 6 min. of lag?
No one would. The entire scientific trove on the other side of that crust will be lost if he twitches wrong.
I’ve got no horse in the AI/ML/neural net race, but this seems like a reasonable task for such things – we would love to be able to place a human in-the-loop on-site, so no lag.
And you are right, currently the next best option is that human-in-the-loop with the lag.
But this _is_ a robotic mission. Giving that robot an additional skill, to be able to train it on a capability it didn’t have when it left earth, by training the local unit’s net then sending the net weights, etc. to the remote device.
There is no joystick; there are commands to move this motor or that motor by this many steps in that direction (maybe move this part of the arm so many millimeters in that direction, but NASA likes low-level interfaces). Every command would be simulated and tested on a replica on Earth before being transmitted.
That kind of verification is the reason you won’t see anybody trusting an operation like this to ML. If they don’t know exactly what it’s going to do, they won’t let it happen. They’re only starting to talk about maybe doing a bit of autonomous navigation on future rovers, and that’s where a mistake only means you have to spend hours manually backing up and going the other way around the rock, not ending a mission with a broken instrument.
More importantly, the lag’s unimportant because when it comes to the rover’s arm positioning, there’s no timescale. It’s not like there’s wind that’ll move the thing or something. Move it a millimeter, take a new picture. Whatever. No big deal. In fact, the only reason the lag’s a pain is because it demands more time on the DSN (and from the Mars relays) and, well, the lander’s got other jobs to do.
The lag was a problem when they were trying to hammer it deeply because there *is* a timescale there – the time it takes to hammer.
“They’re only starting to talk about maybe doing a bit of autonomous navigation on future rovers”
Uh, that’s a bit too far. Curiosity’s been navigating autonomously its entire time on Mars. It does path planning and obstacle avoidance on its own. Opportunity did it before Curiosity, too.
“Its entire time” should say “almost its entire time.” They didn’t do it for a little under a year since obviously they were verifying the rover’s capabilities for a while.
FWIW I’d love to be that engineer.
Using machine learning in such a delicate environment would probably not end well, especially since you can never fully simulate the conditions of Mars here on Earth for training data. It’d probably be just as likely to snap the mole in half as to properly brace it.
Not to mention how risky it is to upload any new executable code to a remote rover, in general.
Whos signs off of these NASA budgets? I’d like to see anyone try to propose a spending spree here on “this” planet and get it passed. They would be the salesman of the year!
Salesman: I need 1.3 Billion dollars for an experiment.
Committee: Why.
Salesman: I would like to drill a hole on Mars five meters into the regolith while dragging a sensor-filled tail behind it, to measure subsurface heat flow.
Committee: For what purpose.
Salesman: Cuz I’M curious.
Committee: sure no problem.
Taxpayers: WHAT THE HELL!!!!
Our schools such, have no money so many other important things are lacking funds, and NASA pissing away money digging holes in the dirt. Just pure genius!!!
But think about where the nasa spending goes. Most of it to salaries, and to suppliers who spend it on salaries, domestic manufacturing, etc. Contrast that to how much money from consumer spending ends up in the hands of foreign owned companies, or frozen away in the cash hoards possessed by companies like Apple.
Typical ignorant response.
First, our schools don’t “have no money”. You may argue that they could use more (and I would agree), you may argue that they could spend what they have more efficiently (and I would agree). But unless you’re going to offer up specific evidence, the blanket claim that schools “have no money” is of zero value.
Second, NASA isn’t just digging holes in Mars just because it sounded like a fun thing to do and maybe it’d get them a bunch of “likes”. They are doing it to expand the human race’s knowledge of the universe at large, and local planets specifically.
Third, NASA’s full budget has been at or below 0.5% of the federal budget for almost the entirety of the past decade, and below 1% for over a quarter century. If you want to discuss the utter waste of federal money, would you care to discuss spending on the military or health care (in a society with a for-profit health care industry)?
Forth, it’s absurd to suggest we put the advance of scientific knowledge on hold until every last human being’s needs, however small, are met to their satisfaction, because there will ALWAYS be someone or some segment of society who feel they need or deserve something.
Lastly, we’re in a race to the bottom with regards to corporate taxes, and we know any savings go towards one, and only one, thing – increasing shareholder value, which predominantly goes to a tiny sliver of the population who’s wealth is already at destructively high levels.
In summary, with all the obvious and egregious examples of federal waste and destructively misdirected monetary policies, to single out a single mission from a government agency who’s budget is less than a rounding error of several other agencies is asinine.
For perspective, there are at least 25 people in the world with a net worth that is greater than NASA’s annual budget.
NASA’s budget 2020: $22.6 billion
https://wealthygorilla.com/top-20-richest-people-world/ (URL says “20” but the list has 25.)
The “poorest” of the 25 is Michael Dell: Net Worth: $31.9 Billion
Michael Dell is an American businessman, investor, philanthropist, and author. He’s the Founder and CEO of Dell Technologies, which quickly became one of the world’s largest technology infrastructure companies.
Richest is Jeff Bezos, who could fund NASA for FIVE YEARS and STILL be a billionaire four times over.
Net Worth: $117.5 Billion
Jeff Bezos is the founder of Amazon, one of the biggest and most popular companies on the web. Amazon started as a simple online bookstore in Jeff’s bedroom, and the initial sales were slow.
We have one of the highest corporate tax rates in the world….Maybe it needs to go down.
You forgot to mention that nobody actually pays those rates, after deductions, exceptions, etc. the actual tax paid (the metric that actually matters) is far less than most countries.
Want to know how to get rich? Contract to teach 5 students at the price public schools charge to degrade their minds. $100,000/year.
Public schools specialize in wasting money, building a bureaucracy, expanding union power, and imparting leftist propaganda. Sending youngsters to public schools is child abuse.
So you’re saying that everyone who can’t afford tuition to a private school should be put to death right now, before they waste any more of our resources Are you going to do your civic duty and commit mass murder?
I know this is wrong. Our local school system has ~7000 students and isn’t getting $700 million/yr. The next county north has 72K students and yet somehow they’re not receiving one sixth of the state budget – in that county alone. There are an awful lot of counties in this state. I think you’re off by about a power of ten. Also, your school experience – something must have gone pretty badly for you. I hate to hear that.
Not that I agree with Chris Maple but….
“Contract to teach 5 students at the price public schools charge to degrade their minds. $100,000/year.”
Looks like he was saying $100k for 5 students, not $100k/student. So that would mean your schools (if at that rate) would be $140m, not $700m (of course I have *no* idea what your schools actually get but you said “off by about a power of ten” but if he meant $20k/student/year, that would bring it down to “off by a factor of 2.” And just to be pedantic a bit :), “off by a power of 10” could mean “the amount was 10^100 times too high/low” as “10^100” is a power of 10. If you meant the amount was 10 times too high (or too low) and not 10^X times too high or low, where X could pretty much be *anything*, the correct wording would be “off by a factor of 10.”)
Even $20K/student is wrong for most locations in the US. The highest per-pupil costs are in the DC area and New York (which, by the way, have many of the best *performing* schools in the country), and those are on the order of $20K/student. Statewide average is probably like $12K/student, and on average only about 60% of that goes to instructors. With an average pupil/teacher ratio of ~16 and a standard overhead cost of ~30-40%, that puts the per-instructor cost somewhere in the neighborhood of $80K/year or so for a 16-person classroom. A 5-person classroom would get you ~$25K, so he’s off by about a factor of 4 or so for most locations in the US. Of course, that’s fully averaged over all grades and all experience levels.
Can’t reply directly to Pat but I did say “if at that rate” :) But it does look like eriklscott’s numbers were probably closer to your $12k than Chris Maple’s (apparent) $20k since eriklscott was saying it seemed 10x too high. I haven’t really looked at the numbers for my area, much less for other areas, so I’m taking your word and eriklscott’s word for what amounts actually are spent and you’re definitely right about how the educator/teacher doesn’t get ALL of it, no matter what the number is.
Education data’s from the US Census. The $12K figure (specifically $12,201) is for 2017, census usually lags by about 2 years. The average going to instructors are just rough (+/-5%-ish), didn’t feel like doing the math.
Overhead’s from BLM. The actual overhead for teachers is a bit higher (~50%) as teacher benefits tend to be a little better, but I was trying to compare it to what you’d get in industry.
https://www.census.gov/newsroom/press-releases/2019/school-spending.html
https://www.bls.gov/news.release/pdf/ecec.pdf
That being said, it’s pretty clear the original statement (“teach 5 kids at cost/pupil and you’ll be rich!”) is very, very far from the truth, even in the highest cost/pupil areas. You’re not getting rich in NYC if your total revenue’s 115K/year.
Not only that, but they put some serious greenhorns on the engineering team for this one.
This experiment is not in the NASA budget. It’s some other government that is paying for this to be there. It’s up to NASA to fix it because the rest of what’s up there belongs to NASA and they aren’t going to let someone else play with their toys.
Out it pops again, and the camera catches a glimpse of a tentacle in the hole – 0_o
What? No Plan B? How tuff would it have been to add a simple rod and knob, like a double headed nail, on the top of the mole so that the scoop, with a notch, could lift the mole, find a new spot and restart the process.
The mole, being of a very clever design is a bad one for this job. They may as well have gone to the local sex toy shop, buy the fanciest marital aid, put a temperature probe in it and sent it to Mars.
The kapton ribbon wire that is attached to the mole seems very fragile and vulnerable. Just a few too many twists or if it is nicked by the scoop or bent a few too many times, it is mission over.
While we are at it, who came up with the idea of that grapple? Did they run out of motors? Is automotive thermostat technology the only thing they could come up with?
If I were going to drill a hole into a planet with unknown properties, I would pay for the extra weight and send a real drill, like a hammer drill with a long hollow drill bit that is NOT open at the end. I would be almost certain that a company like Milwaukee would love to see their name in a picture from the Planet Mars! Once the drill is all the way down then feed the temperature probe (like a two wire thermocouple) down to the bottom. Even if you only go down 4 feet that is far better than the current result of only a few inches.
Think about it, we want to drill a hole into a planet with absolutely no idea what is under the very top layer of sand and dust. There is no doubt in my mind that after hundreds of millions of years, maybe even billions of years, something like layers of sand and clay flower pot like material with maybe some gravel, loose sand or packed regolith, and oh yes, a few rocks. And to all of that let’s add some water, frozen or liquid. Did anyone try this out in very cold salty muddy gravel?
Once the mole starts to go at an angle, then it is just like a nail, your only choice is to pull it out and start a new hole. Woops! Oh, that’s right, we have got no way to retry this! Try as you might, I am afraid this part of the mission is over!
this is comedy gold!
” I would pay for the extra weight” is the plan to take some sort of payroll deduction or are you going to mortgage your house? Are you sure that will be enough money?
You say you have “absolutely no idea what is under the very top layer” and then you apparently have some ideas about “very cold salty muddy gravel” which sounds very specific to me.
You really do have a future in comedy if you can keep it up.
“so that the scoop, with a notch, could lift the mole, find a new spot and restart the process.”
Wouldn’t work. You need the support structure to give the mole enough resistance to get into the ground in the first place. The support structure got moved out of the way, you can’t see it.
Besides… why would moving to a new spot make a difference? The problem’s the *soil*. The soil’s not any different a few feet away.
“The kapton ribbon wire that is attached to the mole seems very fragile and vulnerable. Just a few too many twists or if it is nicked by the scoop or bent a few too many times, it is mission over.”
Why does everyone keep bringing this up? There’s no problem with the FFC. It hasn’t been damaged. It worked fine. The scoop isn’t supposed to be near it at all. The only reason it’s mentioned here is because they’re trying really hard to jerry-rig a goofy solution to an unanticipated problem.
“I would pay for the extra weight and send a real drill”
My God! You should’ve been on the design team! If only they would have thought of sending a real drill! Oh, wait:
“Q: Why doesn’t the mole include a drill?
A: A drill would require a much bigger, more powerful motor than what the InSight lander can accommodate. It would also require more power than the solar-powered lander can practically provide. What’s more, a drill would require rigging to stabilize it as the motor spins, just like with a drill press. Rigging cancels out the force of a drill’s spin, which would otherwise spin the motor in the other direction.”
Plus, to be clear: *this wasn’t the lander’s primary mission*. This is basically an add-on – it never had a super-high chance of success (by NASA standards) even by the project’s own estimation. It’s gotten a ton of attention because it’s having issues (and the XKCD comic) but the real core mission of InSight is basically as a precision weather station, and that part is working *fine*.
Drilling in space has a long history of failure. Even with humans present, only 2 of the 3 drilling attempts actually worked in the Apollo mission, and I don’t believe any other lander has ever managed to do it.
why not try digging a starter hole with the scoop and then putting the mole in said hole? then pack the material removed back into the hole with the mole.
Priceless.
I’m going to ignore all the negativity in the comments here and just say Joe Kim has done it again: Great art!
Yeah, right?
No comment on inter-planetary armchair engineers, other than to say that getting a rover to dig a deep hole, with a sensor package, on Mars, is probably harder than it looks. And it looks hella hard.
Indeed… amid all the negativity… I think that heading picture has nailed it.
Hopefully this doesn’t prompt Warner Bros to come knocking for their royalties for Marvin the Martian’s appearance.
So they built a thumper and sent it to Mars. Good thing there are no sandworms on the planet. https://www.youtube.com/watch?v=72pxx-Fcn_g
If you dig without rhythm…
Interesting that the drill/impact unit is left hand rotation. This means every other thing involved and in contact needs to be left hand also.
If that cam follower is threaded, its direction of rotation and contact with the cam will unscrew a right hand thread.
Then the high angle of taper on the follower (animation) draws my attention, Is that depicted accurately?
IF so, then that’s going to provide a rather heavy (roughly) axial load on the cam follower and tend to pull (or help unscrew) it.
I have to wonder if it has any taper, “in real life”, then why? Are they thinking of directing any burring/chipping aka normal wear damage of the cam , inwards.
So this brings me to the question of scale and proportion of the animation.
In order to understand/speculate any possible/likely points of fouling, You need to know where the clearances/spaces are.
In other words, which direction do you want to try influencing the burr and it be least likely to hang the cam follower and what loading can you influence to best not help “rattle” it apart.
Then we get into material choices and making best guess decisions about assembly materials vs what sort of dust and debris will be encountered by the tooling once it’s actually on the Mars.
and It does seem that any Martian dust/chips they manage to generate may prove to be rather trifling if/when it gets amongst the fiddly bits.
I forgot to ask if anyone else noticed the rate of rotation the borer showed when the scoop was pressing against it?
With hard and slow drilling, how quickly will the umbilical get twisted to a stopping point?
Is there a revolution counter anywhere in the system and does it have some Achilles heel issue waiting for us?
I’m slightly amused that it’s a clicky pen/bathroom pull cord switch mechanism.
I am certainly not an aerospace engineer, but am, nonetheless, dumbfounded that the design used kapton as a cable material. And kapton sheets do not have four degrees of flexibility that a wired cable has. So what am I missing?