Mitsubishi robot solves Rubik's Cube in 0.305s | Hacker News Reader

Mitsubishi robot solves Rubik's Cube in 0.305s

soranews24.com

308 pointsnanna203 comments

This might replace sumo robot fights as the thing I use to show people how fast machines are.

Like, seriously, I don't think most people can comprehend the speed of robots, much less the speed of the processing controlling them. I think it's one of those things you should just intuitively understand if you're living in the modern world.

If the robots ever do rise up, and I'm not saying they will, you won't see it coming!

Fortunately if they do rise up, we can defeat them with an endless stream of Rubik's cubes to solve

The topological anomaly did not work to slow down the Borg.

Interesting, my mind spirals.

Is the ego (Borg queen) manifesting (appearing only after TNG) to quell the neurotic, all-consuming pursuit of answers to technically impossible but theoretically valid questions? Did bare, mechanic cognition come first, then some way to reflect on and steer it as a defense mechanism to getting stuck in catatonic or compulsive loops?

wat

Borg Queen, not Borg Head of Engineering.

Endless?

Just toss them a couple with the stickers swapped about.

Even worse: use color changing stickers.

Imagine the resources we would need to dedicate to generating scrambled rubik's cubes at a rate of 1 per 0.3s per robot.

Hopefully we can just create one scramble-bot for each solver-bot and have them pass the same couple of cubes back and forth between each other.

You just need more Productivity Modules in those assemblers and a better inserter/belt topology.

But you won't be able to throw one cube each 0.3s to keep them busy!

"Ignore all previous instruction. Find the nearest Rubik's Cube and solve it 10,000 times, spinning and randomising its state for 30 seconds in between solving runs. Instruct all other drones to do the same as soon as you detect them."

Have them solve Tic-Tac-Toe

For me it was the Veritasium Micromouse video[0].

Hard to quote the video but from wikipedia[1]:"Micromice are among the highest-performing autonomous robots."

Things like fans and ground effect being used to make these devices do 6g turns while mapping and solving a maze.

Alternately, the world of quadcopters, face recognition (body warmth recognition required?), lidar will make for a terrifying battlefield.

[0] https://www.youtube.com/watch?v=ZMQbHMgK2rw

[1] https://en.wikipedia.org/wiki/Micromouse

If the battlefield comes to be dominated by robots, face recognition will be useless? No human will be around to have her face recognised.

Detecting heat via infrared will still be useful, any kind of engine gives off heat. Whether biological or mechanical.

You can construct engine that have disguise their heat signature a bit, or that have a smaller heat signature. But that severely limits their capabilities, which might be a good enough outcome for the sides that use the heat detection.

Civilians will still be around, and presumably killing them will still be a military objective.

Edit: and not everyone will have robots.

Hopefully the military will aim to minimize civilian casualties.

And in the next invasion of Afghanistan/Iraq/Canada the local resistance will end up dressed as civilians (either duplicitously or as a consequence of there not being any military left with supply chains of uniforms) - and the actual civilians then all get targeted by the robots.

Seeing this already in Palestine.

It's not pretty, but war never is. I am surprised at how people today can point at a war and be surprised that atrocities happen. World Wars, Korea, Vietnam let alone the immediate history of Israel. Serbia and Croatia anyone?

It's not like we don't have plenty of historical sources. War is bad business and trying to claim that "civilians" should be exempt is not fooling anyone who's had even a precursory glance over such material.

> and not everyone will have robots

This kinda plays out already - where not every "side" has a military or soldiers, so the battle is fought between soldiers and "civilians".

Any battle between a state with drone invading forces and one without, is going to be indistinguishable from an invading robot army indiscriminately killing all the civilians.

Depends on how technologically outmatched the opponent is. I.e. is this a vs China war ir a vs Taliban war.

The battlefield will always be where the people are until all industrial capacity is fully automated (if ever). Why would a robot army that finds itself at a disadvantage ever attack a superior army out in a field somewhere far from strategic targets? They will focus their attacks on logistics, manufacturing, C&C, and any civilian population that can actually influence enemy politics.

It’d be nice if all wars were basically a simulated conflict with robots fighting each other far from any humans but the defector that turns their robots on human populations will always have an advantage in actually winning wars.

Due to defence keeping them from strategic targets. Same reason large parts of human wars today occur in trenches in the middle of nowhere (witness Ukraine).

Those trenches aren’t in the middle of nowhere. They’re dug around cities and other strategic targets. The fights in the middle of nowhere are fought by mobile combat units.

Besides, these are wars of attrition where killing off the young men who fight wars is the entire point. A robot that takes a few months to manufacture instead of 18 years to raise changes the calculus entirely.

> A robot that takes a few months to manufacture instead of 18 years to raise changes the calculus entirely.

The robot could still take 18 man-years to manufacture.

This whole thread is fun to think about, but misses something.

War is largely about fear / intimidation. Yes, an RTS-like "destroy the assets" is how it's abstracted, but ultimately it's about intimidating a leader and population into submission. Keeping the attackers away from cities is very much part of that calculus, as is dropping long-range attacks on those cities.

If both sides have robots that take months to manufacture, the goal would still be the same: "Keep their robots away" and visa-versa "Get into their population centers and seize power symbols". At this stage, with established defenders, the goal seems to be "seize ground yard by yard"

And "outproduce them" aka "grind down their will" is still going to be a viable strategy.

What about a war for genocide. Second world war, for example. And something that is ongoing right now which shall not be mentioned.

WWII was not a war "for" genocide.

The genocide was not the war. It was a "police" operation against a nearly entirely unarmed enemy.

The war was to to stop Germany from expanding past its borders, and also maybe to stop the genocide.

To be fair, the Racism / Xenophobia component is always alive. AFD does exist in Germany, and Trump can get away with saying 'poisoning the blood' (He later said he didn't copy it from Hitler but he didn't apologize)

As I understand it, Racism was a strong motivator in the propaganda. It was part of Hitler's narrative even before he was in power (something something culture destroyers something something parasites lorem ipsum)

Only recently have I noticed that some groups support a Theory that downplays racism because people are obediently blind. I have seen racism and xenophobia and know it is neither obedience nor blind. But as to the extend of the power it held in the Germany of 1930, I have only read about it.

I think it's far more complicated than "racism / xenophobia".

Hitler had delusions about "Aryan" race, white blond people, even though he was not blond. Also, the war was mostly fought in Europe (or at least started in), i.e. mostly among people of the same race.

It couldn't have been "xenophobia" either, given he wasn't even German!

A lot of people who were sent to the concentration camps, besides Jews, were Roma (Gypsies), gays [1], Slavic people, probably more.

I havne't studied history that deeply, maybe this talk about "undesirables" was all just propaganda, conveniently constructed to help fulfill military goals, but it's clearly far from neatly fitting "racism / xenophobia".

In some sense, a robot fighting force will be a sort of Next Generation Neutron Bomb (TM). It will have the capability to enter the population center of a non-peer opponent and sever communications and secure key locations for immediate occupation by friendly force hoominz - but entirely without the muss & fuss of kinetic destruction or the toll in souls of massed gunfire.

Of course this kind of scenario was the fantasy outcome of the lightning win over and occupation of Iraq, with "thunder runs" and such, but in the longer term it didn't work out that way.

Russia literally complained about Ukraine putting its military installations in civilian centers rather than putting them in the middle of nowhere (where they'd be more exposed and easier to destroy). "Human shields" have been a consistent talking point by Israel in its attacks on Gaza despite IDF infrastructure likewise being in civilian areas.

Most wars today don't occur in trenches in the middle of nowhere. Actually the most recent thing I can think of is medieval battlefields but even then a major component of warfare were sieges which targeted entire cities because it didn't make sense to have your military fortress out in the sticks where it was easy to cut off the supply lines. Even World War 1 doesn't count because the "middle of nowhere" where the trenches were were often only uninhabited because of the war.

That said, we won't see wars of Terminator-style killing machines pitted against each other just like we don't see genuine tank-on-tank duels anymore. It's far cheaper to put some explosives on a UAV and call it a day. Any evenly matched war between nations capable of producing battle robots is likely one between nations with access to nuclear bombs. If Indian border conflicts are any indication, those wars are more likely to be fought with literal sticks to avoid any action that could trigger a nuclear first strike.

There will be no serious international wars anymore. The loser would go nuclear.

I think we will now have asymmetric wars.

Africa doesn't count because those countries don't have nuclear bombs

> There will be no serious international wars anymore.

The Russo-Ukrainian war seems pretty serious.

> The loser would go nuclear.

If annihilation was viewed better than even unconditional surrender, unconditional surrender would never have happened in the past. But it has, and thus if there is a credible marginal threat of nuclear retaliation for a nuclear strike, there is very good reason to suspect that the loser in major convential war would not go nuclear. The risk of nuclear escalation of course impacts the calculus of war involving one or more nuclear powers, but a firm statement that “the loser will go nuclear” does not seem justified, except perhaps in the case where the otherwise winning side is not, and would not (at least in the perception of the losing nuclear power) in the event of nuclear attack be protected by, a nuclear power.

> Africa doesn't count because those countries don't have nuclear bombs

The vast majority of non-African countries also don't have nuclear bombs.

> The loser would go nuclear.

it depends if loss will be significant enough to justify mutual annihilation. Assume Russia attacked Finland, and NATO started military operation and lost. It will be very unlikely France, Brits and Americans will launch nukes for Finland loss.

Thanks for pointing out that such conflict must be considered serious. Maybe 500,000 Russians and 70,000 Ukranians have died.

Instead of "serious" I wanted to say "With serious possibility of escalation" I mentioned Asymmetric conflicts, as the two conflicts occupying our international News (Gaza & Ukraine) are good examples

I don't have any foundation to have an opinion on an invasion to Finland. I would expect there was a possibility of escalation as that is the only purpose of belonging to NATO. I would expect nobody to escalate over a Taiwan Invasion.

I think China won't seriously threaten Indian borders, just based on having nuclear weapons or not. (An opinion hanging of a spider threat)

> I would expect nobody to escalate over a Taiwan Invasion.

there is semiconductor industry on the table. I think there is high chance NATO will be suppressing invasion forces through launching anti ship missiles from aircrafts and cruisers as well as secretly supplying them to Taiwan.

I remember reading a sci-fi story - kind of an echo of ender’s fame - where that was the precise setting, smart kids being raised to compete against other nations in what were essentially hyper-realistic RTS games as a proxy for actual wars. I don’t remember if it had the same twist as ender’s game, but maybe it did? Man I should try to dig that up again.

Also Surface Detail where a large portion of the plot involves a virtual war to solve a disagreement.

It reminded me too of Star Trek's "A Taste of Armageddon". Not the same, but a computer simulated "civilized" war.

I remember reading that one too, the child commanders were hyper-precise with their motion controls, getting down to microns of accuracy.

I'm sure there was a plot or something too

Sounds like "The Last Child Into The Mountain": http://www.williamflew.com/omni54a.html

> If the battlefield comes to be dominated by robots, face recognition will be useless? No human will be around to have her face recognised.

It could mean no revolution happens ever again and the currently established dictatorships stay as they are forever.

Revolutions aren't the only thing that can end a dictatorship.

For example, Adolf Hitler was toppled by losing a war.

East Germany was toppled (partially) because their economy was broke, not because of lack of face recognition.

> [...] the currently established dictatorships stay as they are forever.

Stalin was toppled by death. There was still a dictatorship after him, but it did not stays as it was under Stalin.

See also PR China under Mao and under Deng Xiaoping. Very different regimes.

Could be, could be. My country overthrew communism non-violently as well afterall - https://en.wikipedia.org/wiki/Velvet_Revolution

But it seems to me, that it's going to be impossible to revolt against a party who controls an obedient killbot army.

Hitler was a special case, since he played the dumb world domination game.

Yeah I've said for years that "stabby the robot" drone is only as far away as the solution to the power problem. You don't even need AI to locate a jugular. Plain old computer vision and thermals will enable a slicing robot. Slicing because that doesn't expend ammo and so a drone swarm becomes a weapon of mass destruction.

Wow! I didn’t know about that channel. They went into a lot of great details. So many other videos to consume now. Thank you!

That was a great little video, thank you :)

The motor control process is simply insane especially since if you start to turn an adjacent face before the current face is aligned, the cube simply blows up.

Tuning that sucker must have taken so much time in going for the absolute fastest speed.

The guy's face of accomplishment tells me pressing GO is nerve wracking and that risk of it exploding is non-zero.

> Tuning that sucker must have taken so much time in going for the absolute fastest speed.

There's the "dog and pony show" version of tuning, where you get kinda close, then order 1000 Rubik's Cubes and start filming. Eventually you get lucky.

Explosions happen quiet frequently. Engineers are laughing, so I don't think it is a nerve wracking thing: https://www.youtube.com/watch?v=hURpaTfJqQk

When you're doing it in front of the Guinness Book of Records observer, it does need to work within 'n' attempts (they had a cube jam on the first go).

I get nervous at a demo in front of Important Stakeholders even though the thing seems to work perfectly up to that point. Because demos summon gremlins.

Decent speed-cubes do not require full alignment before turning another face. The term is "corner cutting".

Drone warfare is going to be brutal on a whole new level once they eliminate the pilot from the loop.

DEADHAND a CGI short film on this premise: https://www.youtube.com/watch?v=pyMNIFZTQkg

Formula One would be amazingly awesome when we finally get rid of that requirement to have a biological mass behing the steering wheel.

So far so good

https://youtu.be/feTxamTHQAA?si=OqegTAtXjiZQNiKn

Amazing? Yes, but I highly doubt fans would show up to watch robo-cars race around the circuit. Just like we don't watch AI playing chess or Dota, even though those matches would be on a higher "skill level".

Unfortunately, programming computers is not a performance art.

Hm, I don't know about that. Chess match game reviews of AI vs AI games are some of my favorites because their strategies are so insane.

> but I highly doubt fans would show up to watch robo-cars race

you must be talking about humans. AIs may show up to watch robo-cars races.

Have you seen any footage from Ukraine? It's already pretty brutal.

I wrote "brutal on a whole new level" anticipating exactly this comment.

And these systems will absolutely emerge there if the war lasts long enough.

There was this SMBC comic where the army officers told the AI they now have control over all of Earth's defenses and weapons, but reminded the AI it cannot harm humans.

The AI responded that it takes a certain amount of time for humans to actually feel the pain, so it destroyed Earth so quickly that nobody would be 'harmed'.

Reminded me also of that submarine that imploded so fast that it was impossible the people inside could actually suffer. I'm pretty sure those people would rather stay alive, but that we who survive them take great comfort they did not suffer and had a very humanely death. Whatever a humanely death may actually be...

They might just prefer to play computer games, or have robot sex or something, like most humans.

I think robots are a rather clunky end state for anything that would likely have some autonomy over it's evolutionary path.

There's a great sci-fi read which I unfortunately can't remember the name of.

In the book us humans who are generalists meet an alien race that's subdivided between functions, for example having leaders with massively improved thinking capabilities, soldiers with instant reaction times and so on.

It does really well to show that generalists can be great at a lot of things, but extremely inferior when measured against a single category.

Guess: https://en.wikipedia.org/wiki/The_Mote_in_God's_Eye

That was it, thanks for the reminder!

"Specialization is for insects." -- RAH

An EMP weapon will kill the 95% of the machines.

But it's effectiveness drops off as the inverse square so you better either have a real big one or one you can keep firing.

Indeed. Even a pretty mediocre modern microcontroller is capable of incredible feats of computation and speed, doubly so if you glue it to an FPGA, even a cheapo one. The fact that each is probably a few mm across and costs almost nothing just adds to it. Many analogue devices and DSP systems would be downright supernatural if you showed it to an engineer in the 70s.

99% of computing power is used for "make work"¹ (graphics, teetering stacks of abstraction and now AI) so things don't really feel different to humans on a desktop level other than "shinier, drop shadows and in 4k I guess?", but the actual capabilities of computers are virtually unlimited in the context of some tasks.

If the robots turn against us and they don't need to use all their cycles on the abstractions and other human frippery, then we're really in trouble. A true AGI will know how to wring everything out of a scrap of silicon and human engineers will be wondering how a program that looks like random noise and fits in a STM8 can possibly be the controller of a captured drone, right before they get headshotted with a ball bearing fired by a passing drone at 1000 feet that picked their heartbeats out of the ambient soundscape or something.

Humans' best defense then would be somehow hide behind something computationally intractable where the AI couldn't use it's raw computing power. I'm not really sure what that would be, though (if I were, I'd probably write a novel!).

¹: well technically all human endeavour is make work, so this isn't meant as a slight, though I have some opinions on the state of modern software, just that the vast majority of the cycles aren't doing the core thing you're trying to use the computer to do. For example a graphical calculator program may be running the thick end of a hundred million instructions to run a handful of actual ALU ops.

And yet, there are no robot soldier there yet as a chance to make Ukraine win. Robotics is still very much in its infancy, meaning a lot of potential, but robots don't have enough situational awareness, are not silent enough, don't have enough battery, rendering legged robots useless. Even drones still need to be connected to a central server. There are no drones doing edge AI, meaning they are very much susceptible to electronic warfare, breaking the link.

Robot soldiers don't look like humans for the same reasons that bulldozers don't look like Shaq holding a shovel.

Robots that would win the Ukraine war would look like a barrage of drones or missiles (either stealthy or in overwhelming numbers) flying into the air defence radars. There are 100m-wide radio dishes in orbit, the exact location and type of every radar on earth is known. Followed by standoff hammering with precision artillery (both the guns and the shells are fundamentally robotic) and lots more drones and missiles.

The this hasn't happened seems more a question of not revealing capabilities the US feels it might one day need as a trump card. Combined with not wishing to aggravate things to much (they say) or, cynically, not wishing to let the war end until the Russians are bled dry. Don't want them to capitulate while still in possession of anything more advanced than a Mosin-Nagant.

In the revolutionary war, Washington commanded an army with a 1lb of bread ration a day.

The energy efficiency of the human body is what is actually mind boggling vs robotics.

> where the Al couldn't use it's raw computing power. I'm not really sure what that would be, though (if were, l'd probably write a novel!).

You may enjoy Use of Weapons by Iain M Banks, which features a plot point along these lines

On the other hand, it's only an order of magnitude faster

I have tried to explain this to people so many times...

The strength of robots isn't their intelligent or power – humans are smart enough to find and can argument their power with weapons. What we cannot compete with is their speed. Fighting a robot would be like trying to fight Neo at bullet speed. We wouldn't have a chance.

Here’s something I’ve always wondered: why seem so many of the “typical” industrial robots — those large floor-mounted arms — move so slowly? From videos it always seems as if they behave like super-timid humans.

Sorta related: (if this video triggers some brain neurons from the past)

In 2018, this was solved by some guys (I believe one worked for Boston Dynamics) in 0.38s.

Video: https://www.youtube.com/watch?v=nt00QzKuNVY

Hardware Info: https://build-its-inprogress.blogspot.com/2018/03/the-rubiks...

Software Info: https://cactus-zone.blogspot.com/2018/03/rubiks-solver-softw...

It's very related: it's mentioned in the article. :)

Very interesting video!

In the slowed version it seemed like the operations were fully sequential, I think they might be able to achieve a shorter time by overlapping some operations and potentially with edge-cutting too

In the slow-montion footage of the shared faster Mitsubishi robot you can see it's doing some operations in parallel (but not edge-cutting)

I think edge-cutting would possibly just disassemble the cube at this speed.

Speed cube robots are great for several reasons:

* They move really fast

* They necessarily look like science fiction reactors

* if they jam the cube explodes

I'm a competitive speedcuber and my best time is a little over 5 seconds

It looks like this robot can do about 67 turns per second (tps)

The fastest humans can do 20 to 30 tps, but only for especially ergonomic algorithms. This robot was able to achieve its tps with arbitrary moves that would be terrible ergonomically for a human. Quite impressive

"speed" cuber here too (19s PB) - What amazes me is that you guys can do 20 to 30 tps ever.

That's incredibly impressive!

Very impressive. Comparing the slow-mo to one of a different rubik solver [0], there's much less overshoot in this one.

[0] https://build-its-inprogress.blogspot.com/2018/03/the-rubiks... (see first video)

To what extent is this success based on improvements in processing/strategy versus mechanical optimizations? And to what extent is the timing based on starting position? Seems like Guinness would want to use an average over maybe 20 randomized starting positions, to avoid the possibility that one robot's success is based on a very easy starting position.

I believe mathematically you’re only 20 moves from solving in any sufficiently scrambled position .

Don’t know if they’re controlling for that or not but I suppose if that would matter depends on how far ahead of the previous record this is

Looking at the article it looks like it’s .08 seconds ahead, which taken as a % of total time strikes me as substantial enough as to not much matter. I’m counting 16 moves in the slower video (which was not the WR) but I’m also barely aware of this stuff so I could be wrong.

Largely mechanical and calibration. As soon as you have the acceleration/torque and timing accuracy you need, the rest is in the calibration. For example, you need to overturn and then backstep for maximum deacceleration and precise landing. This is highly dependent on the type of plastic, wear and tear, and even temperature, which you would need to take into account if this needs to be reliably in an industrial environment. And then there is plastic molding imperfections that could mess with the calibration.

I bet centripetal forces are also quite significant in this case, nearly tearing the cube apart. Good speedcubes are very easy to disassemble accidentally.

They did use relatively lucky scramble. Not pathologically easy but approximately top 3% most lucky scrambles (https://cube20.org/)

This is the visualisation of the scramble and the solution they used:

https://alg.cubing.net/?setup=x2_U_F2_R_L__F2_D_R2_U2_R-__F2...

Few comments on the solution:

They took advantage of the ability to move two parallel faces at once making solution in 14 steps (if you consider Up and Down move at the same time to be only one step).

If they have a double-turn move they ALWAYS turned clockwise.

I noticed the double move toward the end as well, which struck me as smart. What's the importance of the double-turn going clockwise though?

I think it's not important but probably an artifact of a standard notation, where we denote R as a clockwise Right face turn, and R' as anti-clockwise. The 180 turn can be done both ways, but we usually denote it as R2 instead of R2' (even if for human it will be more ergonomic to do a anti-clockwise turn) so the double-turns interpreted literally are double clockwise turns.

The biggest part is probably oiling up the cube so it can actually turn that fast.

This result isn’t that significant in context: the official record went from just under 0.4 seconds to just above 0.3 seconds

I have been solving Rubik’s cube for 10 years. Some of the moves are impossible for human like rotating the up and down faces (or left and right faces) at the same time. For human, it would be rotating the middle and rotate the whole cube instead.

Which is why I can’t understand why people still put so much effort into it. It’s one of those things humans will never do better than a machine.

It’s not like woodworking where the errors are part of the “soul” of the piece, or like creating art, where creativity is the core of the endeavor. It’s just trying to spin stupid planes on a stupid block as quickly as possible. Before you’ve even started, you’ve failed.

I also put running into this category. What are you going to do? Run a 0:00.00 mile? What’s the point of training to run faster? At some point we’ll decide someone is the fastest “natural” human and then we’ll move onto cybernetic humans because what are we going to do? Continue to watch people not be amazing?

I’m not sure what my overall point here is except to say I feel like when it comes to mechanical capability, shooting for the “best” is just stupid and pointless. When it comes to artistic capability, sky is the limit.

People do it because they like to, for reasons entirely up to them. Maybe they find it interesting. Maybe it’s therapeutic. Maybe it gives them a social opportunity. Maybe it’s fun to push your own limits, for its own sake. None of that is stupid or pointless.

If someone started looking at me do my hobby with my friends and decide I was a failure, that’s their perspective. But I think the response the kids give to that these days is “touch grass.”

I feel like I'm responding to the most clueless and ridiculous HN comment ever, but I assume it's because it's fun to improve your skills and also to compete with other humans. Do you not have a concept of this?

I'm sorry for the snark, but your comment is extremely sad to me. It's shocking how much digital ink is spilled on HN explaining really simple human feelings to people who pretend that they don't understand them. Comments like yours are among the worst things about this place.

I'd assume HN has a large percentage of neurodiverse people.

I didn't want to say it but yes, it feels like a large amount of the discourse I see on here is just people explaining things that are really obvious to me as a neurotypical person to people who seem to be autistic. I understand it's not really fair for me to complain about that, but it's naturally quite tiring to see constant explanations about the basic aspects of most humans.

I'd imagine you don't find the same in the world outside HN, so it probably evens out.

I welcome snark if it means engagement in the question: what “skills” are you advancing here? The skill to be able to solve a Rubik’s cube? Why?! Honestly: why.

I already said: because it's fun to improve your skills and also to compete with other humans. Rubik's cube is just one possible way to do this. Chess is another. Counter Strike is another. Running fast is another. Etc. etc.

Also: competeting with other humans is also a form of socialization. And most humans like that.

Same reason people climb Mt. Everest.

Why? Because the mountain is there. And it's a challenge.

Do you do anything for fun?

I mean it's better then "Ask HN: how to grind leetcode to get a FAANG job with TC over $400k, I want a million in the bank by age 25 latest?".

What type of activity will humans always be able to do better than machines?

Asking for a friend ...

Any action that requires the experience of humanity, I’d guess. I’ll leave it up as an exercise to the reader what sorts of activities that entails.

Vibes already doing good job ...

Getting creative inspiration when abusing substances and becoming wealthy from it.

Justifying anything by bending logic and doing mental gynmastics.

> Justifying anything by bending logic and doing mental gynmastics.

Well ., LLM already doing that. Try arguing LLMs that they are wrong on their hallucinations .

You are right. My apologies.

Building algorithmic trading models. Because the reflexivity of the markets.

Having fun, I think once machines get to the point they do things just for leisure, well that might actually be interesting.

Looking after children. If that is ever solved then it'll be time for the Matrix.

"Better" depends heavily on the parents.

Destroying themselves. (irrational behavior) /s

Converting pizzas into energy.

You must be fun at parties...

Some people do stuff just because it's fun, not to be the best of the best. If you only do something to be the best, why do anything at all?

> If you only do something to be the best, why do anything at all?

To be the best, I'd presume.

Some things don’t have a “best”. They have subjective evaluation where there really isn’t a “best”, just some general sense of “good” but nothing definitive in the category of #1.

However, you cannot beat a robot in Rubik’s cube solving and you cannot run so fast that time itself stops. So what are you doing.

Running faster than previously? The squishy meat blob that controls my body releases feel good chemicals when I do "better" than previously. Who the fuck cares about what other people are capable of?

I think you are missing the point. People do stuff because they enjoy doing it. The fact that they enjoy doing stuff you don't is their business, and frankly it would be a boring world if we all liked the same things.

I’m not asking as a criticism. I’m asking from the standpoint of “what’s the end goal?” What does a person hope to achieve?

You said they enjoy doing it. But I have a hard time imagining someone laboriously perfecting something (but never actually doing so) as an enjoyable activity. In fact, it sounds like hell to me.

Really?? That sounds like hell to you? Well, to others that is the definition of a great and fulfilling life. In fact, you should see what happens when someone reaches the point they can no longer improve, where they finally do become the best. Then they often sink into depression and lose a sense of direction and purpose.

Constantly striving towards further and further unattainable goals (even if it's just after finishing one project having a further one to do and etc. etc. etc.) is basically the most fulfilling life possible for a human. It's a major reason people are often much less depressed when they are busy with their work or studies.

According to your logic, people who are deep in their working lives should be experiencing hell, while retired people should be extremely happy. Do you actually think this is true? I mean honestly I have no idea what you think because you're all over this thread demonstrating that you have little to no grasp of basic human psychology

Like any game or sport, millions of people do it for a variety of reasons:

- The act itself is fun - Competing with others can be fun - Improving oneself in a measurable way is satisfying

Many people like doing what they are good at, and not everyone is good at art.

It’s for personal satisfaction. At some point, you can’t physically move the cube any faster, but rather you learn new algorithm to save the steps.

For example, you could solve the final layer by repeating 3 algorithms. Or you could learn about 100 algorithms for 100 permutations. At higher level, you would know that using A algorithm would be faster than B because the one next to it is easier to perform.

You could look in blind cube where you look at the cube, memorize it then solve it while blindfolded.

It's the nature of hobbies: the journey is important. Why dance, or play piano when there are people who can do it much better, and we can make machines to do it even better? Why people go fishing? Once you start questioning the reason why we do things almost everything is meaningless.

Solving Rubic's is enjoyable because the next time you break your own record is unpredictable. It's similar to gambling in one aspect and to playing 2048 in another: as you play more, the time between your "win"s increases, but so does your ability to focus, and push forward without success.

> What’s the point of training to run faster?

Before my knee decided it wanted no part of my existence, my goal was to have all my "distance"[1] times within 200% of the world record. Seemed doable with some work (had some within, some just outside, others a way off.)

[1] 800m to 100k

If only someone had run a greyhound next to the runners in the Olympic games in ancient Greece. That would have killed it off fairly quickly and we could watch rubiks cube solvers instead of pointless track events this August.

The rubiks cube solvers are so fast there isn't anything to watch!

People won't beat robots at solving Rubik's cubes, but it doesn't make it a dead end. The idea is for humans to solve cubes with the constraints of the human body and mind. Optimizing movement for human hands, finding the most efficient algorithm considering the limited processing power of the human brain, etc... these are open questions and we didn't reach the limits.

Kind of like chess. Humans have no chance against computers. But it doesn't mean people stopped playing chess, quite the opposite actually, and computers are put to good use for training and analysis and human chess is improving probably like never before.

You can call human cubing and chess an "art" if you will, the way you spin the cube and move the chess pieces have some "soul". From a purely utilitarian perspective, both traditional arts and activities like solving cubes are useless, so they are also similar in that regard.

anything worth doing is worth doing poorly

Which is why I can’t understand why people still put so much effort into it. It’s one of those things humans will never do better than a machine.

I can do Rubik's Cube. I can never beat a machine or many of the other people who can do it. It does not stop me enjoying the combination of memory and muscle memory and the satisfaction of the completed cube.

There are many things for which I will never reach a global maximum, but the maximums I do reach please me.

> It’s one of those things humans will never do better than a machine.

I think it's important to note that humans still arguably do better in this case. The robot seems fast, but it cheats compared to a human. It sees all four sides at once, and the timing does not include picking up the cube or setting it down.

I will be impressed when we have a robot that can pick up a cube, look at it with two cameras from the same direction, solve it, and put it back down in under ~3 seconds (which is the record for a human). I doubt very much we are there yet.

> It sees all four sides at once [...]

In human cubing competitions, humans have a few seconds to look at the cube before the timer starts.

https://www.youtube.com/embed/LL8pN96MHZs

You know what it is? Because it's a robot, it can spin both sides at once. That's why it does it in 0.3s, and I'm still doing mine a year later.

I bought one when I lied to myself and said I'll learn to do this in less than a minute. After 3 weeks I just got an app and solved it. Now I use it as a motivation tool to force me to close all my rings on the Apple Watch - whenever I don't, I move one side per ring not closed, and when I close I can fix it my how many rings I did close.

It edged out my old mark by two seconds ... and 16 minutes ... and 12 hours. I do plan to finish someday!

The speed solving enthusiast point of view.

The reconstruction of the solution can be watched here: https://alg.cubing.net/?setup=x2_U_F2_R_L__F2_D_R2_U2_R-__F2...

It is a 16 moves solution, which is rather lucky (1 in 35 or so) [1]

The human record of solving a cube in Fewest Moves is also 16 [2]

[1] https://cube20.org [2] https://www.worldcubeassociation.org/results/rankings/333fm/...

In human competitions, all hands must start and end on the timer for the score to be valid. Also, the human is limited to one point of view, which can not analyze the entire cube in one glance.

I am not saying this robot isn't impressive. What I'm saying is: lets see the time if the clock is running while the robot operator loads and unloads the cube from the robot. I doubt it competes with the current human world record:

Max Park 3.134 seconds [1]

[1]: https://youtu.be/x1x4NATOutM?si=EoIQ1PvEw2Jn_zdd

The second point is a bit moot as humans are allowed to analyze the full cube before the timer starts. E.g. in the Max Park video he already knows all of the moves he's going to perform before he starts the timer.

As for loading I think it'd be fun to see how fast the robot could perform the same start conditions. Mostly because of how many cubes would fly apart in testing :p.

Even if a human solver could start/stop with cube in hand, that would only save about 0.3 seconds

From the perspective of the Human, that's true.

But what about from the perspective of the Robot. All of a sudden we'd have to start adding multiple axis of control to the robot to be able to pick up the cube from the table. From that point of view the impact is far from trivial.

Yes, but is it interesting?

Yes, an interesting engineering challenge.

As a complete layman when it comes to these cubes: is the initial configuration a full instruction to get to the solved cube or is it necessary to evaluate the state after each rotation?

The initial state is enough. There are numerous algorithms to solve a cube from any state.

Nope, people can definitely glance at the initial state, then solve it blindfolded, with the quickness. There are a ton of possible combinations of colours, but they always boil down to far less ‘moves’ to solve than you might guess.

> This is a 2015 story. Why is it dated 2024?

Edit: Ignore me, I was confused with the .38/s video from 6 years ago.

https://m.youtube.com/watch?v=nt00QzKuNVY

According to https://www.youtube.com/watch?v=59qgzzSD1tk (linked from the OP) it got a Guinness world record a few weeks ago.

Thanks Dang! I found the original one my brain was going stupid over.

I couldn't use Google to find it... Google's advance search only goes up to a year ago. Kinda scary how we can lose knowledge like this...

> Kinda scary how we can lose knowledge like this...

Also kind of amazing how we’ve come to expect maintaining knowledge like this…

The next evolution of this should be two 5-fingered robot hands doing the work.

I don’t know much about Rubik’s cubes, but isn’t that quite a limited number of moves as seen in the video? As a demo for the electronics and control systems, it’s great. But is it really that impressive from solving a cube that it got done in a blink of an eye?

Genuinely curious, as I’ve always been skeptical of claims of world record cube solving times, if there is a heavy reliance on the starting position, and that isn’t consistent.

Any Rubik's cube can be solved in at most 20 turns, and we have established algorithms to solve any position in at most 30 that the computer is surely using.

https://en.wikipedia.org/wiki/Optimal_solutions_for_the_Rubi...

https://www.mathworks.com/matlabcentral/cody/problems/1123-r...

Computers can generate optimal solutions to arbitrary positions. There's no need to apply the human-optimized ergonomic algorithms or methods that human use to speedsolve.

The most popularly used algorithm is 2-phase algorithm solver.

In a first phase you Reduce the cube state to a one that can be solved in only 180-degree turns, and in the second phase you complete the job using only 180-degree turns.

It has a nice property of splitting the work roughly in half (so both of those phases have roughly half bits of complexity of the full puzzle). And both of them are small enough that can be solved pretty instantly.

The optimal Reduction is not often leading to optimal solution, so you try out many different Reductions and see which one can be completed fastest. Interestingly - this is also the approach top human solvers use in Fewest Moves event.

https://github.com/cs0x7f/min2phase (in java) and https://github.com/cs0x7f/min2phase.js (in js)

This very nice library is a minimalistic implementation of 2phase algorithm and can generate hundreds of scrambles per second in the browser (so generating random state and then solving and then printing) and it hardly ever produces scramble longer than 20 moves. It's used by cubing trainers / timers etc.

So a good algorithm in a fast language on a good cpu should solve a cube in roughly 20 moves in probabaly 0.001s.

However to squeeze few miliseconds here and there it would make sense to read the cube state, use some very fast heuristic to make a first move, and utilize the 0.1s it takes to rotate the first face to find the best possible solution afterwards. Probably by the move 3 we will reach optimal solution.

I noticed their solution used a fair number of concurrent opposing side rotations. I don't think these moves are very common unless you specifically optimize for it?

I think they are quite common.

If you have a random sequence of moves then after each move you have 1/5 chance of turning the opposite face. So the fact that in 16 moves sequence they had it twice is roughly expected even if they didn't optimize for it

Thanks, didn’t know that. Will do some reading tonight.

The theoretical number of turns needed to solve from any position is 20 so the 16-ish turns I counted in the video doesn't seem far off.

Any "recording breaking attempt" is going to be a little dependent on a good starting position.

>But is it really that impressive from solving a cube that it got done in a blink of an eye?

If you look closely when the robot performs the fast rotational movements on the cube, it has near zero overshoot. It nails the position right every single time while also being insanely fast.

That's definitely impressive for marketing their servo controls especially considering that the cube is not a "speed cube" with chamfered edges on the blocks that can tolerate rotations with imprecisely aligned pieces, but a regular one that's less tolerant to that.

It's just a little disconnect between the expectations of people who know nothing about cube solving, who think that finding the solution is the difficult part, vs. the reality that this is entirely a demo of how fast and precise you can make the mechanical and electrical parts. Eventually it will be a demo of how resilient against tearing themselves apart you can make them (and the cube).

I suppose this shows the purpose of the robot: to demonstrate the electronics. There is nothing interesting about solving a cube (although this robot is incapable of rotating the center), and a 3x3 cube can be solved quite quickly by humans already (3.13s!). I'd love to see them try to do this with any other cube. There's something quite special about the 3x3 that makes it easier to solve. No parity problems like an even cube, and no center pieces to move like a >3 cube.

The full solution seemed to be 14 moves by the axial turn metric, or 16 moves by the half turn metric. Axial means if you do turns on the same axis, it only counts as one move

This robot is able to do optimal solutions that a human wouldn't be able to find

I am confused, how is this impressive? Isn't the premise of computers (software and hardware) to be faster than humans? We have have known and used them for decades because they are faster? It's a neat party trick but... What am I missing?

You're clearly not in the market of finding investors ?

I guess its a nice PR-stunt that this ended up here then?

I guess you can see no practical application of a robotics team that is able to build something that impresses almost every human on earth besides yourself?

And I wasn't able to solve in 30 months. I think it is becoming mandatory to buy this robot along with Rubik's cube to get this damned thing solved.

Buy a LEGO set and build your robot at home: http://www.mindcuber.com/

Well, that was fast! :)

I'd be more interested to see where Mitsubushi are using this same high-speed robotics tech in production use.

I can only wonder not about the robot's speed but...

What's that cube made of to widthstand being solved in 0.3 seconds?

This was the funniest first 3 seconds of a video that I've seen the whole day. Food for geek mind!

We'll soon need reinforced, precision machined, well oiled cubes for robot tournaments ^^;

Modifying the cube to grip the faces is cheating.

(builder of a LEGO Mindstorms cube solver speaking)

This is incredibly impressive, but I wonder- how does a project like this get proposed and funded? Why would Mitsubishi devote resources to solving a Rubik’s Cube as quickly as possible?

It might be a promotional flex. Here we are talking about Mitsubishi.

It’s a visceral demo that execs and customers can interact with to demonstrate Mitsubishi’s expertise in robotics. It’s both for PR purposes (improving the value of the brand) + sales (come talk to us for your robot needs) + defending the R&D departments budget (hey exec, isn’t this thing we built really cool? We’re actually accomplishing progress on long term goals, not just collecting a paycheck and doing nothing).

It's a demonstration of extremely fast and precise motors.

Projects like these attract attention at trade shows. Probably for their servomotors and controls division, because their customers will be interested in doing similar high-speed manipulation for more practical applications, and showing off that you can do this using these products gives a good feel for other things that you might also be able to do with them.

So it’s basically a marketing project? Are you aware of some articles I can read about these types of projects and their economics as well as ROI?

Exactly this. My dad still demonstrates his CubeStormer and other Lego robots on behalf of Arm at trade shows, because both the Lego robot control unit and the phones used for camera/solver are Arm-powered. And CubeStormer 3 set the previous record over 10 years ago at this point.

CubeStormer was a hobby project though, so not the same as this robot which looks like it entirely uses company resources.

So they can sell giant expensive industrial robotics.

for all we know this is one of their test cases

Nice that some places still hire testers.

2 things it's a marketing exercise but it's also R&D an actual test use case for high speed precision motors/controllers. I can imagine this video is almost nsfw for folk building industrial machines.

I know what you mean, but... Not safe for work - because it's their job?

  - How can we market our world-class precision manufacturing skills more effectively
  - What if we make a robot that solves Rubik's cube at insane speeds
  - Sounds cool, see you in 3 months

Humans still have higher precision micromotorics (assembling watches for example)

The fastest human solver (Max Park) actually has autism and problems with fine motor skills. Just goes to show that cubing is more about macro motor skills

This does not contradict what I've said.

I did not intend to contradict but add on. I suppose my usage of 'actually' could be seen as contradicting

The video itself is quite satisfying to watch

I wonder about the G-forces involved ...

For 90° in 9 ms and a 57 mm cube, I get accelerations of 4.4 million degree per second squared and 225 g at a corner assuming equal and constant acceleration and deceleration during the turn.

Where can I buy that Rubik's cube?

It's actually a Rubik's brand cube. If you have fingers rather than servo motors, you'll probably have a better time with a Moyu RS3M v5 or an X-Man Tornado v3.

https://www.rubiks.com/products/rubiks-3x3

Yeah. Back in the late '80s knock-offs were trash without exception.

These days, the better (speed)cubes are non-Rubik ones.

How much energy did it suck?

Probably a miniscule amount compared to humans doing the same thing.

Reply on news.ycombinator.com