Back in December 2020, Hyundai purchased an 80% controlling interest in Boston Dynamics from SoftBank for $880 million, part of a transaction that valued the robotics company at $1.1 billion. That agreement included a put option allowing SoftBank to sell its remaining stake to Hyundai at a later date.
Oh. It's Softbank exiting humanoid robotics at Softbank's discretion. That's a lot different than " Hyundai buys Boston Dynamics". Hyundai bought them years ago. This is just the last 8%.
I don't understand why they would implement humanoid robots instead of purpose-built robots. The human form is not the most optimal way to do most tasks, especially as it relates to manufacturing. Robots don't need to look like humans, they need to be useful. Seems like putting in an awful lot of extra unnecessary work to end up with a worse result.
I'm not sure how many times this has to be restated.
It's car manufacturing. Everything that could be done by a purpose specific robot arm bolted down to the factory floor is already done by a purpose specific robot arm bolted down to the factory floor.
What remains unautomated, then?
The long tail of tasks that are too minor, too finicky, too open-ended or too reliant on manual dexterity to be offloaded onto traditional robots.
This is where this new generation of robotics comes in. This is the kind of task they're designed to do: "a task that's still done by a human in a high automation environment". Universal robots are angling for the tasks that are impossible or uneconomical to automate with traditional industrial robots.
> Everything that could be done by a purpose specific robot arm bolted down to the factory floor is already done by a purpose specific robot arm bolted down to the factory floor.
Hah! Hardly. I say this as someone whose first "real job" was in applying robotics research to automotive assembly - there are still a ton of assembly tasks that could be performed by a fixed-base robot arm, or a robot arm on a linear rail/fixed gantry. Wheeled mobile manipulators are only needed in a few cases, and humanoid form-factor is only "necessary" in very few cases (and I don't think the current crop of humanoids is particularly suited to these tasks).
In my opinion/experience, the impediments are that (1) the system integrators that are usually responsible for assembly-line robotics are too stupid to figure out how to apply robots to the problem, (2) the automakers themselves are often too short-sighted/stupid/unwilling to invest in increased automation (and particularly in building the in-house competency that they really need), (3) the hostile/exploitative relationship between (most) automakers and their main suppliers means that low-hanging improvements to parts/assemblies are a non-starter, and (4) the automaker C-suite (and investors) are too drawn to silver-bullet solutions (e.g. humanoids) than practical automation improvements.
"Could be in principle" and "could be in practice, under technical and economical considerations in play" are two very, very different beasts.
Everyone in the industry learned that the hard way.
At a certain point, the tasks that remain stop being "dexterity" problems and start being "AI" problems. That is: a robot could do the task - if you either spent big $$$ on redesigning the entire task around the robot's intellectual limitations (uneconomical), or if you had an incredibly advanced AI capable of problem solving driving that robot (impossible with 00s AI).
The "universal robot" bet is the "incredibly advanced AI capable of problem solving" bet. That in 2020s, AI is finally capable. The body only has to be "good enough to make most tasks possible".
> "Could be in principle" and "could be in practice, under technical and economical considerations in play" are two very, very different beasts.
> Everyone in the industry learned that the hard way.
The auto industry is notorious for making incredibly myopic choices to save money/make money in the near term versus long-term investments. The relationship between automakers and their suppliers/vendors is basically a century-plus of the automakers trying to (1) outsource anything they can for a quick buck, and (2) grind the supplier/vendor margins down to nothing. (This is part of why the newer Chinese automakers with much greater vertical integration are such a threat to the traditional automakers; vertical integration has a high up-front investment but the payoff in flexibility and speed is significant).
I have visited factories for work and my experience is the same. There is so much stuff that could easily be automated but is not because it is too expansive for too little value to make a custom one off machine. The big high volume things will be automated but these machines will have 90% success rate and lot's of stuff that needs to be done by hand. You can search for factory tours on youtube to get an idea. Here are two videos, an Amazon warehouse and a Tesla factory. the big heavy stuff is automated but lot's of work is still done by hand.
https://www.youtube.com/shorts/-R6cBkza17khttps://www.youtube.com/watch?v=45slYC99uUg
This is over the last decade at one of the largest automakers in the world. Naturally there is significant variation between individual lines and plants; some are newer and more automated, some are older and much less automated. Are some cars being built on more automated lines? Yes. But a great many, probably the vast majority, are being built with fairly low assembly* automation.
* There is a significant split in automation between "body weld" stages and "assembly" stages. Body weld is very heavily automated basically everywhere (although there are some surprising exceptions in places), while assembly is much less automated.
“One of the largest automakers in the world” makes me think of very low-tech companies like Ford or whatever. I can’t imagine this would bring much actual experience with this new generation of robotics.
automotive workers unions started around 1918 and became major political players in the 30s -- a fact that i'm sure is wholly unrelated to why there are so many un-automated tasks in that industry.
But if these tasks are too minor, too finicky, too open-ended or too reliant on manual dexterity for a purpose-built robot, how can a general purpose robot perform them better? If anything, they should be doing worse.
The only thing I can think of are tasks that are so rarely done, it's not economical to build a robot for. But I then I also don't see how another robot solves this problem.
A) the idea is that these robots do have dexterity capabilities a lot closer to human hands
B) there’s a long tail of individual tasks it’s uneconomical to build purpose-built robots for each individual task. But it’s economical to have 1 robot that can do all of them.
Do they? A human can both chuck kilograms of stuff across a room or kick in a door, but then pick up a single hair off the ground, or feel and manipulate (things even lighter than) a literal feather.
Robots can certainly do things more repeatably, if not more precisely.
The human arms and hands are very versatile, and imitating them is a good choice for a universal robot, though 3 or 4 arms are definitely better than 2, and the hardest to imitate are the sensors, not the actuators.
But the rest of the human body is not useful in a factory environment, so the arms could be mounted on a mobile base that does not have any resemblance to a human.
And C) they don't always have to be at parity with human hands to be good enough because humans are flat out expensive. Humans need health accommodations, have sick days, vacation, and make mistakes too. The bar is much lower and the incentives are much higher than many people probably think.
These robots operate on completely different principles.
One can lift insane weights, has insane torque, and absurd precision, and can do the same movement millions of times with virtually no deviation. You program these with an exact movement plan, just like you would programm a CnC with a tool path. They are basically cnc machines.
The other one is a inacurate, unstable, dynamic system controlled by neural networks and heuristics. It has massive deviation over each run, but that means that the programming must be able to account for it. Which makes it suitable to operate on problems that are messy, unrepeatable and human-shaped.
Pretty much. It's a total paradigm shift from how industrial robots normally work. A pre-planned motion executed carefully and precisely vs open-ended "do this thing" powered by a very large bag of opaque neural network heuristics.
A robot that has to be carefully adapted and set up for the task vs a robot that you can point at a task and have it figure out how to do it.
It's a bet that The Bitter Lesson will win over Moravec's Paradox, in the end.
It's also worth noting that when e.g. inputs to a stage might have unpredictable defects or alignment, a robot arm utilizing neural networks for planning and analysis might still be the best way to handle that - without the extra degree of freedom of movement-relative-to-floor, planning can be done more rapidly, and movement can be executed more aggressively and quickly.
If I were Hyundai, I'd be looking at this as buying a significant amount of vision, dynamics, and integration systems expertise, not necessarily the dream of self-motive walking systems.
Well, humans obviously do those jobs, so a clearly a general purpose robot (in this case, a biorobot) has been found to do the job better. Don't overthink it.
Because it is general purpose. We did not have the ability to create a single robot form which could do all of these minor, finicky, and opened ended tasks. Now that seems within reach. The nice property of humanoid robots is that the world is already made for human form, and so if you're trying to replace people naturally this is what you'd want.
>> how can a general purpose robot perform them better
Better than what? It seems that as long as they perform the tasks "better" (cheaper / faster / lower-error) than the humans that are currently performing them, that is an improvement for the factory owner.
It's not a "general purpose" robot, it is a "human replacement" robot, with similar skills and shortcomings to a human. Humans are not general purpose.
All you need to do is look at a recent video of car manufacturing process, and watch what the humans are doing.
>It's car manufacturing. Everything that could be done by a purpose specific robot arm bolted down to the factory floor is already done by a purpose specific robot arm bolted down to the factory floor.
>What remains unautomated, then?
Stuff that can be done by purpose specific robot arms on wheeled platforms, which is very difficult, but will be much more feasible than a humanoid robot doing anything.
What you need then is a better arm (or even just hand), not a human.
Or a new take on car design with automated production in mind regarding all the wiring and what not (easier said than done, I'm sure many have tried and failed, but eventually someone will succeed).
Naw, the real answer is that factories have been built around human labor - they weren't built to be forward-compatible with purpose-built robots, so during the transitional period where we build these purpose-built robots, you need humanoid robots to fill in the parts where the factories were geared for human labor.
I took a tour of the BMW Group Plant Spartanburg body shop. It's heavily automated with industrial robots inside safety cages. But they still have human workers pick up parts from rolling carts and place them into templates for the robot arms. BMW has been running a trial with Figure humanoid robots to automate that remaining piece. Apparently those robots haven't worked very well, but presumably Hyundai thinks they can do it better?
Some colleague of mine visited the Mini plant earlier this week. Apparently they had a Boston Dynamics dog patrolling simply to spot stuff that had been left where it wasn't supposed to be left
Humanoidal robots make sense when they need to operate in spaces designed for humans bodies. Cars are designed and built to be used and serviced by humans, especially their interiors so you need humanoid robot to automate building them. Car exteriors are not built for humans to interact with so they are already being built by specialized robots.
Custom built robots are expensive (basically an R&D project in themselves) and inflexible, if you want to update the process you have to redesign your automated system. The dream of humanoid robots is they can adapt to new manufacturing processes like human workers.
I understand the sentiment, but you assume they are planning to build humanoid robots to walk around the human-oriented space in the factory.
Perhaps they want to put some of the sensing and control features in, so a humanoid-like dexterity or adaptability for the business end of a floor-mounted robot arm?
Training data of task completion. See, e.g., robots doing backflips. Presumably there’s an optimal robot for gymnastics but if you start with humanoid form you can train based on many videos of human movement. The alternative - world model sim with physics and loss functions- perhaps ends up being too unconstrained when you add in optimization of the robot form…
Because a humanoid robot can replace (theoretically) any human worker, whereas a purpose-built robot can only replace one kind of worker. Or at least that's what Asimov said in Caves of Steel.
The gist of it is that all tools on the spacecraft (eg: space-drill, space-coffee-maker, space-airlock) are all designed to fit a gloved human astronaut hand. Waaaaay more complicated to make a robo-hand than a robo-suction cup or robo-claw, but then you are matching the environment, and guarantee tool compatibility against all extant tasks!
We already have specialized robots on earth... paper slicer, lawn mower, bazooka, whatever. They're all machines that are specialized for the task at hand, we're not making a humanoid robot that gets down on all fours and individually plucks blades of glass.
The car factories already have specialized robots... they're not mimicking a human hand holding a can of spray paint, shaking it up, and painting the car that way... it's a 6-axis arm, or a whole "grab the car and flip it while spraying paint" system.
It's not about inventing purpose-specific robots, it's about handling that long-tail of "stuff with tools that a human is designed to be able to use." Go over there, push that button. Go move that box from table1 to table2. Etc.
For well defined tasks in the factory domain, make a "real robot". For ad-hoc tasks in the interim... strap an LLM to a camera, battery, robo-legs and arms, cross your fingers, and hope for the best?
I'm spitballing here as I don't actually have a concrete answer for you. But from my understanding automobile manufacturing is one of if not THE most advanced 'purpose-built' robotics sectors. While I agree with you that having a purpose-built thing usually wins out for assembly line manufacturing, I wonder if this isn't an attempt to branch out away from single-purpose robotics into more general or multi-faceted manufacturing.
I think the rationale is that they are already using typical car factory automation, but they see a huge potential market for general purpose robotics in the coming decades, they don't need the humanoids, they are simply dogfooding a future product.
I think this is smart and not very risky. Tesla is playing a similar game with Optimus, for now Hyundai/Boston Dynamics is at least 5 years ahead.
That's like saying there's no need for a generic CPU. The only way forward is a ASIC. Once a generic CPU does everything well enough, it's extremely versatile.
They have god knows how manny bajillion dollars tied up in machinery designed for human use, a human can step in when the robot breaks, and they can buy more robots if the humans get uppity. Those seem like a bunch of good reasons to me.
The humanoid form factor is certainly may not be ideal but I guess they think the flexibility is worth it
You don’t understand because you’re not an expert? First off they have numerous non-humanoid robots if you follow them at all. Second, Clearly they’ve got strong reasoning, they’ve just been bought. Third, out of thousands of attempts at humanoids, their robots are seemingly the best we’ve yet seen in this class. They must be doing it right when so few others got any traction.
This same nonsensical question again. Because the world is built for humans, because these are general purpose to replace a human laborer. It can immediately go from picking up parts in a factory to mowing the lawn in the same day.
I don't think this is solely tied to car manufacturing automation. Even though Hyundai Motor Group is acquiring them, I would imagine they'd be well-positioned to commercialize general-purpose robotics and not just for car manufacturing, if Tesla is anything to go by.
I do think this might be tied to South Korea's demographics, by 2040 the working-age population is projected to decline 25% from 2020 and keep declining almost linearly until leveling out around 17M around 2065, a 50% drop total in < 50 years. I would think HMG / Hyundai sees a huge business opportunity or this might be a national-level political priority but I don't know the specifics.
Hyundai bumped their ownership up to 100%, and took the opportunity to reset expectations about when Atlas would be working in Hyundai factories.
While Atlas is the best humanoid robot so far, it still isn't useful in a car factory that's fully equipped with the latest factory robots that are strong enough to juggle car engines and are bolted to the floor.
Everyone knows the killer app for humanoid robots is building the Mars colony amirte?
There are plenty of tasks a relatively weak humanoid robot is well suited for. Basically any task that is 'human shaped' and not worth an automated line.
There's got to be something wrong at the core of BD. They've been pawned off a bunch of times, and they still don't have products out the factory line like they should. I think the tech community has been impressed by their videos, but the fact that their most sold thing is a toy dog at a luxury car price point says a lot about the company.
My personal take is that one of the reasons is their posture against ML. They've been very "GOFCT" and have only recently started to incorporate ML concepts.
>> There's got to be something wrong at the core of BD. They've been pawned off a bunch of times,
Well...there is the uncanny similarity to the T-800 and and uneasy realization that the owner of BD could become Cyberdyne Systems IRL. Perhaps some companies like that notoriety but not sure if many want that.
What is Gofct and does robotics industry generally just have had a slower adoption of ML because of the realtime domain requirements, I'm just curious and wondering aloud here.
Is it really that surprising that no-one has invested the time and effort into figuring out the personal information of some tech employee-turned-founder? I bet no-one outside of tech even knows his name.
Yeah. Google was too impatient and forced BD to productize prematurely (Spot, Handle), then dumped them when it didn't work out immediately. AI just wasn't ready yet. Imagine if Google had let BD focus on research until DeepMind was ready with the AI side of things. I think with the right joint research program they could have already been deploying humanoids today.
ai generated imagery can’t be copyrighted while all other photography can and generally needs to be treated as it is. Therefore you likely have to pay a royalty to Getty or other asset outlet. Of use AI.
There's a saying in Zen which I live by. "How you do something is how you do everything".
Start to be sloppy somewhere, you'll be sloppy everywhere. As we "learn and enable" to do things faster with less effort, the quality of the thing we (as in humans collectively) do decline.
AI, when used as the sole blunt instrument, accelerates this dramatically.
Little appreciated fact is news orgs have full time employees just dealing with licensing all day long, and they pay out millions of dollars when someone fucks up.
So true. I hear they replace more engines than any other brand. I'm surprised they sell so well; a used Toyota would be a far better choice than a new Hyundai
Although confusingly, the battery/motor on a Hyundai EV is covered under a different 10-year/100K warranty which does transfer to used car buyers. Important because of their unfixed ICCU problems.
Their much hyped 10 year 100K mile power train warranty is only for the original purchaser. Once sold the warranty reverts to a more standard 5/60K term.
Of the important life-lessons to have before one turns 18;
"don't ever be the third or fourth owner of a Hyundai" is right up there with not eating the yellow snow.
There's no shame in being broke, of course-it is merely a catastrophe. The fourth owner of a '11 Sonata is gonna have a different outlook than the fourth owner of a '73 Mercedes 600 Pullman.
So this appears to mean that Hyundai is effectively taking BD's humanoids "off the market" (B2C/B2B markets). And Softbank wants to take a different humanoids stake in OpenAI's plans.
Just a note for the thread Hyundai Motor Group makes cars and a lot of other heavy industry things - trains, defense, plant. Rolling robotics fully into their motor group makes complete sense.
rnteresting part is defense although the economics and power/range limits its practicality.
the other timing here is the increasingly belligerent union who are demanding pretty outrageous compensation for what a typical American worker would make. I think the goal is to immediately automate the workforce and move the plants out of korea speaking to insiders.
This is a bit disappointing, isn't it? Boston Dynamics had the coolest robots and everybody was marveling how they would take over the world eventually. Turns out the market isn't gigantic and the use cases are limited, at least for now.
However, let's hope they will keep on doing cool stuff under their new owner.
Outside of some military applications and maybe search and rescue, a lot of people kind of freaked out about Boston Dynamics. They have cool robots, sure, but at what cost if they are implemented by a bad actor? No thanks.
I don't think that follows. Hyundai could well sell these after they've dogfooded them for a while.
Car factories seem to be a pretty good initial market, given that Tesla is doing Optimus and Figure has humanoids in a BMW factory. But the whole point is that these are general purpose robots, and there are lots of other factories. By the time that market is saturated they'll be capable of more.
Back in December 2020, Hyundai purchased an 80% controlling interest in Boston Dynamics from SoftBank for $880 million, part of a transaction that valued the robotics company at $1.1 billion. That agreement included a put option allowing SoftBank to sell its remaining stake to Hyundai at a later date.
SoftBank has now exercised that option.
Oh. It's Softbank exiting humanoid robotics at Softbank's discretion. That's a lot different than " Hyundai buys Boston Dynamics". Hyundai bought them years ago. This is just the last 8%.
That feels so low of a price when compared to the insane valuation people attribute to Tesla robots
Softbank is bleeding money and they need cash, AI isn't shaping up to what they thought it would be
I don't understand why they would implement humanoid robots instead of purpose-built robots. The human form is not the most optimal way to do most tasks, especially as it relates to manufacturing. Robots don't need to look like humans, they need to be useful. Seems like putting in an awful lot of extra unnecessary work to end up with a worse result.
I'm not sure how many times this has to be restated.
It's car manufacturing. Everything that could be done by a purpose specific robot arm bolted down to the factory floor is already done by a purpose specific robot arm bolted down to the factory floor.
What remains unautomated, then?
The long tail of tasks that are too minor, too finicky, too open-ended or too reliant on manual dexterity to be offloaded onto traditional robots.
This is where this new generation of robotics comes in. This is the kind of task they're designed to do: "a task that's still done by a human in a high automation environment". Universal robots are angling for the tasks that are impossible or uneconomical to automate with traditional industrial robots.
> Everything that could be done by a purpose specific robot arm bolted down to the factory floor is already done by a purpose specific robot arm bolted down to the factory floor.
Hah! Hardly. I say this as someone whose first "real job" was in applying robotics research to automotive assembly - there are still a ton of assembly tasks that could be performed by a fixed-base robot arm, or a robot arm on a linear rail/fixed gantry. Wheeled mobile manipulators are only needed in a few cases, and humanoid form-factor is only "necessary" in very few cases (and I don't think the current crop of humanoids is particularly suited to these tasks).
In my opinion/experience, the impediments are that (1) the system integrators that are usually responsible for assembly-line robotics are too stupid to figure out how to apply robots to the problem, (2) the automakers themselves are often too short-sighted/stupid/unwilling to invest in increased automation (and particularly in building the in-house competency that they really need), (3) the hostile/exploitative relationship between (most) automakers and their main suppliers means that low-hanging improvements to parts/assemblies are a non-starter, and (4) the automaker C-suite (and investors) are too drawn to silver-bullet solutions (e.g. humanoids) than practical automation improvements.
"Could be in principle" and "could be in practice, under technical and economical considerations in play" are two very, very different beasts.
Everyone in the industry learned that the hard way.
At a certain point, the tasks that remain stop being "dexterity" problems and start being "AI" problems. That is: a robot could do the task - if you either spent big $$$ on redesigning the entire task around the robot's intellectual limitations (uneconomical), or if you had an incredibly advanced AI capable of problem solving driving that robot (impossible with 00s AI).
The "universal robot" bet is the "incredibly advanced AI capable of problem solving" bet. That in 2020s, AI is finally capable. The body only has to be "good enough to make most tasks possible".
> "Could be in principle" and "could be in practice, under technical and economical considerations in play" are two very, very different beasts. > Everyone in the industry learned that the hard way.
The auto industry is notorious for making incredibly myopic choices to save money/make money in the near term versus long-term investments. The relationship between automakers and their suppliers/vendors is basically a century-plus of the automakers trying to (1) outsource anything they can for a quick buck, and (2) grind the supplier/vendor margins down to nothing. (This is part of why the newer Chinese automakers with much greater vertical integration are such a threat to the traditional automakers; vertical integration has a high up-front investment but the payoff in flexibility and speed is significant).
How long ago was your robotics experience?
An Amazon warehouse or Tesla factory tour would likely change your mind.
I had to do both of these in the last year and not a lot of humans around…
I have visited factories for work and my experience is the same. There is so much stuff that could easily be automated but is not because it is too expansive for too little value to make a custom one off machine. The big high volume things will be automated but these machines will have 90% success rate and lot's of stuff that needs to be done by hand. You can search for factory tours on youtube to get an idea. Here are two videos, an Amazon warehouse and a Tesla factory. the big heavy stuff is automated but lot's of work is still done by hand. https://www.youtube.com/shorts/-R6cBkza17k https://www.youtube.com/watch?v=45slYC99uUg
> How long ago was your robotics experience?
This is over the last decade at one of the largest automakers in the world. Naturally there is significant variation between individual lines and plants; some are newer and more automated, some are older and much less automated. Are some cars being built on more automated lines? Yes. But a great many, probably the vast majority, are being built with fairly low assembly* automation.
* There is a significant split in automation between "body weld" stages and "assembly" stages. Body weld is very heavily automated basically everywhere (although there are some surprising exceptions in places), while assembly is much less automated.
“One of the largest automakers in the world” makes me think of very low-tech companies like Ford or whatever. I can’t imagine this would bring much actual experience with this new generation of robotics.
automotive workers unions started around 1918 and became major political players in the 30s -- a fact that i'm sure is wholly unrelated to why there are so many un-automated tasks in that industry.
Hyundai’s manufacturing facilities in the U.S. are not unionized.
Amazon warehouses still have a huge number of ununionized workers doing manual labor
Also, a general purpose robot vs a custom purpose robot represent very different capital investment profile for the factories?
But if these tasks are too minor, too finicky, too open-ended or too reliant on manual dexterity for a purpose-built robot, how can a general purpose robot perform them better? If anything, they should be doing worse.
The only thing I can think of are tasks that are so rarely done, it's not economical to build a robot for. But I then I also don't see how another robot solves this problem.
A) the idea is that these robots do have dexterity capabilities a lot closer to human hands
B) there’s a long tail of individual tasks it’s uneconomical to build purpose-built robots for each individual task. But it’s economical to have 1 robot that can do all of them.
These industrial robots have much better dexterity than any human alive.
The point is, human shape plus general purpose intelligence is an amazing combination to resolve the “long tail”.
Without the intelligence part, the body is useless.
Perhaps Boston Dynamics has that part resolved now too.
> much better dexterity than any human
Do they? A human can both chuck kilograms of stuff across a room or kick in a door, but then pick up a single hair off the ground, or feel and manipulate (things even lighter than) a literal feather.
Robots can certainly do things more repeatably, if not more precisely.
Intelligence is absolutely a valuable addition to dexterity, but no, current industrial robots have nowhere near the dexterity of a human hand.
The human arms and hands are very versatile, and imitating them is a good choice for a universal robot, though 3 or 4 arms are definitely better than 2, and the hardest to imitate are the sensors, not the actuators.
But the rest of the human body is not useful in a factory environment, so the arms could be mounted on a mobile base that does not have any resemblance to a human.
And C) they don't always have to be at parity with human hands to be good enough because humans are flat out expensive. Humans need health accommodations, have sick days, vacation, and make mistakes too. The bar is much lower and the incentives are much higher than many people probably think.
and humans collude via unions
These robots operate on completely different principles.
One can lift insane weights, has insane torque, and absurd precision, and can do the same movement millions of times with virtually no deviation. You program these with an exact movement plan, just like you would programm a CnC with a tool path. They are basically cnc machines.
The other one is a inacurate, unstable, dynamic system controlled by neural networks and heuristics. It has massive deviation over each run, but that means that the programming must be able to account for it. Which makes it suitable to operate on problems that are messy, unrepeatable and human-shaped.
Pretty much. It's a total paradigm shift from how industrial robots normally work. A pre-planned motion executed carefully and precisely vs open-ended "do this thing" powered by a very large bag of opaque neural network heuristics.
A robot that has to be carefully adapted and set up for the task vs a robot that you can point at a task and have it figure out how to do it.
It's a bet that The Bitter Lesson will win over Moravec's Paradox, in the end.
It's also worth noting that when e.g. inputs to a stage might have unpredictable defects or alignment, a robot arm utilizing neural networks for planning and analysis might still be the best way to handle that - without the extra degree of freedom of movement-relative-to-floor, planning can be done more rapidly, and movement can be executed more aggressively and quickly.
If I were Hyundai, I'd be looking at this as buying a significant amount of vision, dynamics, and integration systems expertise, not necessarily the dream of self-motive walking systems.
Well, humans obviously do those jobs, so a clearly a general purpose robot (in this case, a biorobot) has been found to do the job better. Don't overthink it.
Because it is general purpose. We did not have the ability to create a single robot form which could do all of these minor, finicky, and opened ended tasks. Now that seems within reach. The nice property of humanoid robots is that the world is already made for human form, and so if you're trying to replace people naturally this is what you'd want.
>> how can a general purpose robot perform them better
Better than what? It seems that as long as they perform the tasks "better" (cheaper / faster / lower-error) than the humans that are currently performing them, that is an improvement for the factory owner.
It's not a "general purpose" robot, it is a "human replacement" robot, with similar skills and shortcomings to a human. Humans are not general purpose.
All you need to do is look at a recent video of car manufacturing process, and watch what the humans are doing.
>It's car manufacturing. Everything that could be done by a purpose specific robot arm bolted down to the factory floor is already done by a purpose specific robot arm bolted down to the factory floor.
>What remains unautomated, then?
Stuff that can be done by purpose specific robot arms on wheeled platforms, which is very difficult, but will be much more feasible than a humanoid robot doing anything.
What you need then is a better arm (or even just hand), not a human.
Or a new take on car design with automated production in mind regarding all the wiring and what not (easier said than done, I'm sure many have tried and failed, but eventually someone will succeed).
> I'm not sure how many times this has to be restated.
This strike me more as a repeated internet myth more than anything else. There is near endless opportunity for purpose-specific robot forms.
> What remains unautomated, then?
And the tasks that change from day to day.
Naw, the real answer is that factories have been built around human labor - they weren't built to be forward-compatible with purpose-built robots, so during the transitional period where we build these purpose-built robots, you need humanoid robots to fill in the parts where the factories were geared for human labor.
They're just one of today's lucky ten thousand.
https://xkcd.com/1053/
According to this widely cited comic strip, if you are over 30 and didn't know it, you should be ashamed.
[delayed]
I don't think that's what it's saying at all. It explicitly says we should not be negative towards people for not knowing things.
The math on the xkcd is just wrong tbh. In multiple ways.
I took a tour of the BMW Group Plant Spartanburg body shop. It's heavily automated with industrial robots inside safety cages. But they still have human workers pick up parts from rolling carts and place them into templates for the robot arms. BMW has been running a trial with Figure humanoid robots to automate that remaining piece. Apparently those robots haven't worked very well, but presumably Hyundai thinks they can do it better?
https://www.bmwgroup.com/en/news/general/2024/humanoid-robot...
Some colleague of mine visited the Mini plant earlier this week. Apparently they had a Boston Dynamics dog patrolling simply to spot stuff that had been left where it wasn't supposed to be left
BD was training Atlas for Hyundai factory in Savannah, Georgia for a while:
https://www.youtube.com/watch?v=CbHeh7qwils
Can't you just, you know, stick another robot arm on the damn carts and have it offload itself? Surely there's a simpler way.
Humanoidal robots make sense when they need to operate in spaces designed for humans bodies. Cars are designed and built to be used and serviced by humans, especially their interiors so you need humanoid robot to automate building them. Car exteriors are not built for humans to interact with so they are already being built by specialized robots.
Custom built robots are expensive (basically an R&D project in themselves) and inflexible, if you want to update the process you have to redesign your automated system. The dream of humanoid robots is they can adapt to new manufacturing processes like human workers.
I understand the sentiment, but you assume they are planning to build humanoid robots to walk around the human-oriented space in the factory.
Perhaps they want to put some of the sensing and control features in, so a humanoid-like dexterity or adaptability for the business end of a floor-mounted robot arm?
Training data of task completion. See, e.g., robots doing backflips. Presumably there’s an optimal robot for gymnastics but if you start with humanoid form you can train based on many videos of human movement. The alternative - world model sim with physics and loss functions- perhaps ends up being too unconstrained when you add in optimization of the robot form…
Two things: 1) we have abundant training data for humanoid embodiments (watch humans do things), and 2) the world is already designed for humans.
Maybe it's like Formula 1: it's a purposefully-extreme goal, which drives new development and makes "lesser" goals feel more achievable.
Because a humanoid robot can replace (theoretically) any human worker, whereas a purpose-built robot can only replace one kind of worker. Or at least that's what Asimov said in Caves of Steel.
I'm looking for a better video of it (from one of the engineers), but look at the NASA robot hand. https://www.youtube.com/watch?v=vfDXzkFHnz0
https://www.google.com/search?q=nasa+robonaut+video+hand+why...
The gist of it is that all tools on the spacecraft (eg: space-drill, space-coffee-maker, space-airlock) are all designed to fit a gloved human astronaut hand. Waaaaay more complicated to make a robo-hand than a robo-suction cup or robo-claw, but then you are matching the environment, and guarantee tool compatibility against all extant tasks!
We already have specialized robots on earth... paper slicer, lawn mower, bazooka, whatever. They're all machines that are specialized for the task at hand, we're not making a humanoid robot that gets down on all fours and individually plucks blades of glass.
The car factories already have specialized robots... they're not mimicking a human hand holding a can of spray paint, shaking it up, and painting the car that way... it's a 6-axis arm, or a whole "grab the car and flip it while spraying paint" system.
It's not about inventing purpose-specific robots, it's about handling that long-tail of "stuff with tools that a human is designed to be able to use." Go over there, push that button. Go move that box from table1 to table2. Etc.
For well defined tasks in the factory domain, make a "real robot". For ad-hoc tasks in the interim... strap an LLM to a camera, battery, robo-legs and arms, cross your fingers, and hope for the best?
I'm spitballing here as I don't actually have a concrete answer for you. But from my understanding automobile manufacturing is one of if not THE most advanced 'purpose-built' robotics sectors. While I agree with you that having a purpose-built thing usually wins out for assembly line manufacturing, I wonder if this isn't an attempt to branch out away from single-purpose robotics into more general or multi-faceted manufacturing.
I think the rationale is that they are already using typical car factory automation, but they see a huge potential market for general purpose robotics in the coming decades, they don't need the humanoids, they are simply dogfooding a future product.
I think this is smart and not very risky. Tesla is playing a similar game with Optimus, for now Hyundai/Boston Dynamics is at least 5 years ahead.
Our'legacy' world is built for human shsped operation. So a generic operator will hsve to mimic human shape.
The human form IS the most optimal way to do most tasks
The first somewhat general purpose humanoid robot will sell like gangbusters to the rich, even just as a parlor toy/trick.
That's like saying there's no need for a generic CPU. The only way forward is a ASIC. Once a generic CPU does everything well enough, it's extremely versatile.
They have god knows how manny bajillion dollars tied up in machinery designed for human use, a human can step in when the robot breaks, and they can buy more robots if the humans get uppity. Those seem like a bunch of good reasons to me.
The humanoid form factor is certainly may not be ideal but I guess they think the flexibility is worth it
Patient complex moves first.
You don’t understand because you’re not an expert? First off they have numerous non-humanoid robots if you follow them at all. Second, Clearly they’ve got strong reasoning, they’ve just been bought. Third, out of thousands of attempts at humanoids, their robots are seemingly the best we’ve yet seen in this class. They must be doing it right when so few others got any traction.
This same nonsensical question again. Because the world is built for humans, because these are general purpose to replace a human laborer. It can immediately go from picking up parts in a factory to mowing the lawn in the same day.
> they need to be useful
They would be. When everything what could be done would be done by a robot. 24/7. Even without the lights on the floor.
Why do you need software when FPGA can solve everything.
It's much cheaper to create one general-purpose robot for everything than many different robots, each optimized to each task.
Robot Chicken had a fairly cynical take on this. I won't link it here.
I don't think this is solely tied to car manufacturing automation. Even though Hyundai Motor Group is acquiring them, I would imagine they'd be well-positioned to commercialize general-purpose robotics and not just for car manufacturing, if Tesla is anything to go by.
I do think this might be tied to South Korea's demographics, by 2040 the working-age population is projected to decline 25% from 2020 and keep declining almost linearly until leveling out around 17M around 2065, a 50% drop total in < 50 years. I would think HMG / Hyundai sees a huge business opportunity or this might be a national-level political priority but I don't know the specifics.
Wait, haven’t they already owned them for years? Edit: right, they’re just buying the remaining 9%.
Hyundai bumped their ownership up to 100%, and took the opportunity to reset expectations about when Atlas would be working in Hyundai factories.
While Atlas is the best humanoid robot so far, it still isn't useful in a car factory that's fully equipped with the latest factory robots that are strong enough to juggle car engines and are bolted to the floor.
Everyone knows the killer app for humanoid robots is building the Mars colony amirte?
There are plenty of tasks a relatively weak humanoid robot is well suited for. Basically any task that is 'human shaped' and not worth an automated line.
I still think dumping BD was one of the biggest mistakes of Sundar's career. And that's saying something.
There's got to be something wrong at the core of BD. They've been pawned off a bunch of times, and they still don't have products out the factory line like they should. I think the tech community has been impressed by their videos, but the fact that their most sold thing is a toy dog at a luxury car price point says a lot about the company.
My personal take is that one of the reasons is their posture against ML. They've been very "GOFCT" and have only recently started to incorporate ML concepts.
>> There's got to be something wrong at the core of BD. They've been pawned off a bunch of times,
Well...there is the uncanny similarity to the T-800 and and uneasy realization that the owner of BD could become Cyberdyne Systems IRL. Perhaps some companies like that notoriety but not sure if many want that.
https://terminator.fandom.com/wiki/T-800
What is Gofct and does robotics industry generally just have had a slower adoption of ML because of the realtime domain requirements, I'm just curious and wondering aloud here.
Sorry, I wanted to make a pun for GOFAI (good old fashioned AI). CT stands for control theory.
Per this sale, BD is worth $3.25B. Just recently, Google paid $2.7B for two years of Noam Shazeer through the Character.ai deal.
This seems like a small correction if they wanted to reacquire and clearly the market isn't valuing BD all that high.
Why do you think it's one of Sundar's biggest mistake?
https://en.wikipedia.org/wiki/Noam_Shazeer
Can we go on a small tangent and wonder how we don't know when the guy was born?
Is it really that surprising that no-one has invested the time and effort into figuring out the personal information of some tech employee-turned-founder? I bet no-one outside of tech even knows his name.
Yeah. Google was too impatient and forced BD to productize prematurely (Spot, Handle), then dumped them when it didn't work out immediately. AI just wasn't ready yet. Imagine if Google had let BD focus on research until DeepMind was ready with the AI side of things. I think with the right joint research program they could have already been deploying humanoids today.
Google dumping a project when it does not produce instant results?
That seems out of character
It’s so weird to use an AI generated image for this article when there are so many images of Atlas out there.
Take a step back and look at this article's diction and the rest of this entire website. Completely AI generated.
All those tokens have to go somewhere
Generating an image from an already open tab is faster than making a search engine query and selecting a good, high resolution image.
Who cares about quality. Speed is the new black.
Letting AI generate your image also dances around the issues of attribution and licensing, for better or worse.
ai generated imagery can’t be copyrighted while all other photography can and generally needs to be treated as it is. Therefore you likely have to pay a royalty to Getty or other asset outlet. Of use AI.
…who have quite conveniently already stolen and trained on all the copyrighted images. Thanks AI!
Please. It's all fair use. Otherwise AI companies can't repackage and sell what's out there for free.
On a relevant note: https://www.theverge.com/news/674366/nick-clegg-uk-ai-artist...
and tech companies wonder why consumers hate AI
What seems to be the problem here? Why is it offensive that someone didnt spend more time hand selecting a picture for the article?
There's a saying in Zen which I live by. "How you do something is how you do everything".
Start to be sloppy somewhere, you'll be sloppy everywhere. As we "learn and enable" to do things faster with less effort, the quality of the thing we (as in humans collectively) do decline.
AI, when used as the sole blunt instrument, accelerates this dramatically.
The text reads LLM-ish as well.
And figure out usage and licensing and all that crap. So much easier to just generate a brand new image.
Little appreciated fact is news orgs have full time employees just dealing with licensing all day long, and they pay out millions of dollars when someone fucks up.
Yeah, now they don’t need that department.
If only they could make an engine or transmission that doesn't blow up at 80,000 miles.
So true. I hear they replace more engines than any other brand. I'm surprised they sell so well; a used Toyota would be a far better choice than a new Hyundai
Luckily they have a 10 year 100k mile warranty.
...which AFAIK isn't transferable. IOW, used Hyundais are cheap for a reason.
Are you sure? The warranty is on the car, not the owner. Almost all manufacturers (except Tesla) transfer automatically and are based on mileage.
You have it backwards. Hyundai, KIA, etc will knock it down to 5yrs/60K for used car buyers.
Source: https://www.kbb.com/car-advice/car-warranty-guide/
Teslas new car warranty transfers as is to the new owner.
Source: https://www.tesla.com/sites/default/files/blog_attachments/m...
Although confusingly, the battery/motor on a Hyundai EV is covered under a different 10-year/100K warranty which does transfer to used car buyers. Important because of their unfixed ICCU problems.
5yrs is only for the 2nd owner. 3rd owner and beyond gets the shaft. Or rod, I suppose.
Why would they do that? It just seems like such a bad business move without really affecting anything.
Some MBAs probably calculated that it saves more money than they would lose by pissing off used car buyers. And they want people to buy new cars.
> And they want people to buy new cars.
Seems like the effect would be to take off a significant chunk of value off your used car in a way that makes it more difficult to buy a new car.
Their much hyped 10 year 100K mile power train warranty is only for the original purchaser. Once sold the warranty reverts to a more standard 5/60K term.
Who told you that? Unless you buy your Hyundai new, or CPO from a dealership, the powertrain warranty is 5/50.
Only if you happen to be the second owner. Third owners get nothing.
Of the important life-lessons to have before one turns 18; "don't ever be the third or fourth owner of a Hyundai" is right up there with not eating the yellow snow.
There's no shame in being broke, of course-it is merely a catastrophe. The fourth owner of a '11 Sonata is gonna have a different outlook than the fourth owner of a '73 Mercedes 600 Pullman.
Hyundai EVs are great though, the Hyundai Ionic 5N is probably the best EV there is (for car enthusiasts).
So this appears to mean that Hyundai is effectively taking BD's humanoids "off the market" (B2C/B2B markets). And Softbank wants to take a different humanoids stake in OpenAI's plans.
Hyundai could be the first owner that can actually turn BD's robotics in real product
BD always seemed more like an R&D company to me or even a university lab. Doesn’t seem like a good fit for a car company.
Just a note for the thread Hyundai Motor Group makes cars and a lot of other heavy industry things - trains, defense, plant. Rolling robotics fully into their motor group makes complete sense.
rnteresting part is defense although the economics and power/range limits its practicality.
the other timing here is the increasingly belligerent union who are demanding pretty outrageous compensation for what a typical American worker would make. I think the goal is to immediately automate the workforce and move the plants out of korea speaking to insiders.
`buys` sounds like they have just purchased BD, should be `takes full control` or something similar.
I’m pretty surprised to see no mention of Agility in the conversation about other humanoid companies
Boston Dynamics gets passed around again
Nope, Hyundai already owned it, they’re just going to 100%
Title really should clarify: Hyundai takes full control of Boston Dynamics
Rest is previously reported stuff.
Related from earlier in the year:
Hyundai Introduces Its Next-Gen Atlas Robot at CES 2026 [video]
https://news.ycombinator.com/item?id=46520508
And old discussions when the intial buy happened:
2020: Hyundai to acquire Boston Dynamics
https://news.ycombinator.com/item?id=25363981
2021: Hyundai acquires controlling stake in Boston Dynamics for $880M
https://news.ycombinator.com/item?id=27588047
BD always felt like they had some awkward demos but never really revolutionized anything. Now they seem far behind Chinese companies. What happened?
nice
This is a bit disappointing, isn't it? Boston Dynamics had the coolest robots and everybody was marveling how they would take over the world eventually. Turns out the market isn't gigantic and the use cases are limited, at least for now.
However, let's hope they will keep on doing cool stuff under their new owner.
Outside of some military applications and maybe search and rescue, a lot of people kind of freaked out about Boston Dynamics. They have cool robots, sure, but at what cost if they are implemented by a bad actor? No thanks.
I don't think that follows. Hyundai could well sell these after they've dogfooded them for a while.
Car factories seem to be a pretty good initial market, given that Tesla is doing Optimus and Figure has humanoids in a BMW factory. But the whole point is that these are general purpose robots, and there are lots of other factories. By the time that market is saturated they'll be capable of more.
If it's disappointing then it's been disappointing for over 5 years, since Hyundai has owned the company for 5 years.