spot dog is hydraulically powered junk, unitree is motor driven from day one. Boston Dynamics was forced to switch to a motor driven architecture after it is proven by unitree.
Perhaps because of the potentially slower actuation speed, but you also generally get a lot more power from hydraulics so im not sure one can claim it is junk. Far less acrobatic, but also far more sumo wrestler.
Some people on X are saying they're "just" cloning/copying "puppet" human movements.
I know very little about robotics, but given these appear totally free-standing, if that was the case (I personally don't think it is), wouldn't that imply they have the same centre of gravity and weight of limbs as humans? Surely they'd have to be able to balance themselves, and copying a human's movements "exactly" wouldn't work for their own motion otherwise?
I think when watching I saw one or two of the robots "judder" their feet a bit out of sync with others - this seems to imply they are capable of balancing their own motion a bit individually.
I've worked on much less expensive, much smaller humanoid robots.
These robots are certainly running through a scripted set of poses which has been extensively tested for the conditions (Humans would also be choreographed and have to hit certain marks at certain times). If you covered the stage in loose gravel or a thick carpet they'd all start falling over. The things the robots hold are almost certainly taped into their hands.
Despite that, this is a very impressive demo. Those robots are $40k+, they've got 20+ of them. And not a single one fell over. They're fast too - and there are a load of corners they could have cut, but they didn't.
The floor has two textures, it would have been easier without that. The humans right alongside them? Much less safety paperwork without them. The robot wearing trousers and a cape? Much easier without that. The fewer robots you have, the lower the chances on falls over landing their backflip. Lose the audience and record it in multiple takes. Hell, you could have human acrobats in robot costumes and it'd cost far less and be much easier.
So this demo is very much a costly signal of confidence.
It’s not a 1:1 human motion capture to servo translation. There is some work done to fix Center of gravity like you said and issues with friction and momentum.
The hard part with “autonomy” is interpretation of the environment and feeding that back into some control loop to accomplish a goal in real time. That is why most of these demos are basically recordings of movements, like choreography.
They're also interacting with the environment (vaulting boxes / walls), which implies they either know their 3D position very accurately, or they have some form of sensors and can adapt a bit.
Of course the robots have been pre-trained and the movements are scripted, and nobody is claiming otherwise. But there must be a lot of autonomous balancing taking place. At one point you can see the robots adjusting their feet slightly different although they are all in sync, and that catapult does not look like its movement is exactly the same every time. It is just super impressive.
Does anyone remember when Honda's Asimo robot clumsily fell down the stairs during a demonstration[1] and we thought we were safe from a robot invasion by just moving to the upper floor? That was about 20 years ago.
The impressive part here isn't the movement itself. You can easily train a model to perform a "procedural animation" that includes a full body control policy. The hard part is making it reliable enough to perform long sequences of movements and adapting to differences in robot placement. In other words, performing a flawless stage play is the hardest part.
I'm afraid you might not understand what you're talking about. Animation is a geometry problem, while robotics is a dynamics problem. The latter is subject to constraints many times greater than the former. There is no such "easy" model as you imagined that can transform the former into the latter.
It's still a completely farcical attitude. At this point it's just a matter of how many years it'll take for chinese manufacturing to outstrip western expertise with chips, too. Ten years? Five? Two?
Yes, the autonomy level of these robots was what I was yesterday emailing with my former colleagues we were wondering. Two months ago CNET & PC-Mag posted following video which suggests more about robots movements being assisted by humans. And it also shows Chinese have being edge of the development at that point.
There is another match viewable by pressing that "Robot plays ping ppng #robot" arrow.
How about that robot? Is it human assisted or not? Our opinions diverted, I'm quite sure it is assisted but my former colleague thinks it's got to be autonomous as it would be too difficult and slow to do that fast movements with remote control assisted robot.
It would be nice to hear opinions about that playing robot too if anyone could provide some insight in that.
edit: I think the serve waiting robot hand movement and after losing wiping left eye gesture as a disappointing a bit in my opinion gives up it's human. Or if not, why would a robot do such a human like gestures.
edit2: OK, good points, I see now. It's definitely a fake. Thanks to all who replied :)
that ping pong video is a CG robot, whether realtime superimposed or otherwise who knows. Look at the :27 when it gets out of tracking breaking all of physics, feet aren't planted to the ground, light, shadows.. etc.
I think the ping pong match video might be misleading you. Based on the visual artifacts around the robot, the original footage likely had a human player that was swapped in with a robot in this video. It also has an altered content warning.
I can think that future use of pingpot robot is to replicate specific pro player style (from various recording) and be used to spar by pro players before their specific matches.
The second video you've linked is fake in every aspect in regards to the robot.
The robot is floating above the ground.
The paddle is phasing in and out of existence.
The robot has a realistic human hand and uses it to hit the ball.
The robot randomly turns around mid-air near the end of the video.
The robot looks nothing like a Unitree robot.
Oh, how could I forget, the entire robot looks so obviously fake even when disregarding all of the above that I can't believe you're even trying to analyze anything in that video.
I'm 99% sure that ping pong match is CGI. The whole robot has this green screen effect. Look at its feet. And at second 17 it just disappears entirely for a few frames.
I was born in 1989. The most impressive sudden technological advance I have experienced have been LLMs. This video is a good candidate for second place. I am mindblown... That they even dare having children dance with them. The trust they must place. An acquantance bought a chess board with a robot arm, and it accidentally broke his finger because he picked up a piece that the robot arm wanted to pick up. China isn't just a few hours in the future, more like decades it feels like.
Last time I had to implement a typed programming language, it was for short programs only. So for typing I just converted the program into a large set of constraints and throw z4 at it, asking to optimize for smallest possible types. If z4 could find types for every expressions in less than a few seconds then it was well typed.
You had to use a few trick for larger programs, but basically it managed to type any real programs and I never encountered an ambihuous case that caused a problem in practice. In case of failure, the small set of unsatisfiable constraints was easy to translate into nice error messages. This also allows for typing rules that are easy to state and can accomodate operators that adapts nicely to their environment.
I would understand if this approach would be frowned upon, but I still wonder if any serious language ever used this approach?
We are definitely on an exponential in term of capabilities of humanoid robots.
We are probably only years away from having a robot in the house, in construction of robots. Automating anything that a human can do is best done in a human sized robot.
But.
None of these are actually useful right now. I don't want something with the arm strength of a forklift taking care of my parents or kids. The demand for humanoid robots right now is like lift a fridge from a delivery truck to a house (aka more mobile forklift) or walk through toxic sewage to pull crates out. Super useful but basically just mobile cranes, which is a small market. China seems to be making the mistake of pushing a tech demo as a consumer product (we've all been on those projects...) which can make people hate the tech.
Build something people want, don't mandate what they want. We're like 3-4 generations from amazing, useful robots. I'll be scared when these things are minding a bunch of dogs on stage.
> I don't want something with the arm strength of a forklift taking care of my parents or kids.
A risk I never hear discussed is falling over and injuring children. Even the petite Unitree models are like a 70kg piece of furniture. Each year thousands get injured because of furniture falling over. I'd buy one immediately, but if I had kids or pets, I would wait for safety data on falls.
One big issue is the "joints". It's always a trade-off between mass, strength, speed, precision, and dexterity. State of the art is matching or exceeding organic joints on 2-3 axis, while being an order of magnitude below human/animal performance on the two other axis.
Human individual fingers can withstand internal loads of hundreds of newtons, (possibly a thousand for brief periods if you're a star rock-climber), while at the same being capable of tasks such as writing, which require high speeds and (sub-)millimeter precision, while also enjoying 4 degrees of freedom (5 for the thumb), and they're stuffed full of sensory organs to boot. Oh and they're also self-healing, so taking some damage during use is no problem and they will actually adapt to the task they're used for over time. Everything we can make compared to this is laughably primitive.
If you make an agile robot big, it now weighs more, which means its joints now need to handle greater loads, which makes it necessary to make them much bulkier and heavier.
A lot of this is just inherent limitations of electric motors and having to convert rotational energy. This gets heavy fast.
Decent synthetic electrically-driven muscle fibers would go a long way.
So Atlas can lift 7x the capacity. Even Digit, the tote-consolidating robot, can do 35lbs.
Unitree's demos are a lot of fun, and the antics of releasing the G1 to the public has certainly captured people's attention, but a "working" robot won't look, act, or develop from the G1 or even H2.
I wasn't trying to say that Unitree is somehow deficient. I'm sure they could build Atlas if they wanted.
My point was that BD could probably build a robot with the shown acrobatic capabilities, but they choose not to because their goal is to build robots that carry heavy loads for industrial applications.
The point is that a robot with higher load capacity is necessarily less agile. BD's target market is industrial so their robots are necessarily larger and less agile.
The fact that Unitree's robots appear so acrobatic reflects that they are likely on par with BD in terms of capabilities but have a different target market.
Boston dynamics is far behind plus the robots are so cheap , even their dog is cheaper than BD. I dont think their humanoid can even catch up to this price. I am sure US Army and for the chinese counterpart Chinese army will be their biggest customers. But i wonder how will this workout in situations like Plane hijack, fire fighting and other such places where human lives cant be risked to save more human lives. (Please Dont downvote because your american patriotism is poked try replying.)
What really stood out was that when they portrayed different important jobs it was all done by men and women were in the background as decoration/onlookers in awe. Very strange development.
I'm not so sure what this question means... are you asking if China has AGI right now? I'm quite sure all similar performances done by humans are all staged.
Where human performances are staged, the human still needs be capable of learning the choreography. There's a difference between a robot learning the routine vs having it hardwired.
Based on where other companies are right now, it's probably a pre programmed routine, but the robots autonomously balance themselves. Anything beyond that would be quite a large leap, and I think Unitree would've gloated about it way more outside of the gala. The robots' speed and consistency are still impressive, though.
What is the most impressive is the robustness. Of course they are following a captured human routine, but they are facing so many disturbances from which they need to recover and keep following the desired trajectories, while under multiple constraints (movement ranges, not losing balance, etc).
You can see on the backflips that all robots landed quite differently, some with both knees on the ground, some with one, some with none. Yet all recovered gracefully and moved on to the next step of the choreography.
It is genuinely impressive, and scary.
Meanwhile in the west we are bickering like 10year olds.
Oh god the bickering. For self driving cars, lidar vs cameras is totally missing the point. Waymo can drive with cameras only. The AGI question is what decisions does it make when things go wrong.
one interesting observation is that none of those companies are located in Shenzhen, which arguably has the best supply chain for all electronics stuff. I guess those trillions $ spent on infrastructure paid off - Shenzhen didn't suck all talents into its proximity, it becomes an enabler for industries across the country.
Agibot is not in Shenzhen?
I think what happened is that other major cities/provinces like Hangzhou, Suzhou, Shanghai or Hefei started bankrolling talents and enterprises so they too can have major high tech enterprises in their town
AGIBot is the poster boy of the municipal government of Shanghai.
> major cities/provinces like Hangzhou, Suzhou, Shanghai or Hefei started bankrolling talents and enterprises so they too can have major high tech enterprises in their town
there is reason why in those 5 minutes you see them together in same shot with audience only 2 times and only for few seconds
there is reason why most of the shots are not wide angle showing whole scene, seems they learned their lesson from last year where you could easily see on the edges all the failures
this was heavily edited and repeated, I mean is it really surprising considering all CGI you see during whole gala? I watched whole 5 hours (though skimmed through a lot), they just can't make show same as seen by real people on the site, what you see in TV is very different from what audience has seen
edit: the whole gala show is recording, it would be impossible to organize such event across many cities with so many performances live, olympics opening ceremony is walk in the park compared to this
While I don't know whether this was indeed broadcast live, at least this recording is missing a section since as the YT comments point out at 1:25 the staffs appear out of nowhere.
I was initially somewhat skeptical as well because this looks like a surprisingly massive leap in robotics capability, but haven't been able to find anything particularly sus.
Ahhh! When he said staffs I assumed he meant people, not staves! Yeah, that's 100% CG. Here [1] you can see where the kids take a stance where it was to be edited in, and if you pause at the exact moment - it looks pretty bad, with something like a lightsaber just popping in the kid's hand.
Hmmm. Not sure what to make of this. I wish it was possible to see the raw unedited footage. It makes me question everything else.
If you watch at 0.25 speed, you can see the staff is actually a magician's cane. It's a coiled up bit of plastic that springs open when you release the ends. You can see similar tricks on YouTube, like https://www.youtube.com/watch?v=d-j6G4sPwWU
Have you seen the full gala? There are many sections where they interlace full intense cgi, it’s kinda like the new apple conference transitions where they purposefully don’t aim for reality, more like an enhanced experience that may or not have a purely live version.
Yeah, I noticed the heavy editing of the gala this year too and it was very disappointing. Even in tons of performances / dances where they really shouldn't have needed to there were obvious cinematic shots were not 'live' from the same recording. While for the robots I can imagine the pressure might have lead to editing a few takes, it really took away from the regular dancing performances and made it feel a bit more like watching a heavily edited music video.
1) Cool, but when are they actually going to drive my car for me?
2) Any semblance of American technological superiority is pure fantasy at this point. The only area where Americans are truly "advanced" is in selling overpriced SaaS products. There are dozens of Chinese startups with robots just like this—as seen at CES—yet Boston Dynamics is still treated like it’s some untouchable, DARPA-level tech.
3) A lot of this comes down to cost: you can either hire one American fresh grad or a Chinese PhD for the same price.
3) The second reason is cultural: Americans tend to buy solutions, while the Chinese prefer to build them. Even SMEs in China maintain internal dev teams to build custom software for the business, as opposed to paying Salesforce for what is essentially a glorified Excel sheet with sprinkles of automation.
4) America is facing its own innovator's dilemma. The country is currently being run by MBAs and salespeople focused on extracting every last dollar from the consumer instead of providing real value or innovating. Perhaps we're one step beyond the innovators dilemma. The innovators are dead and we are in the corporate greed stage.
5) Americans are completely oblivious to how advanced China has become because of the propaganda they're fed. My personal "aha" moment was when Chinese EVs hit my local market and completely obliterated legacy automakers on both features and price. The American "free" (lol) market is being guarded by politicians but that won't work for long.
#4 is the biggest problem, by an overwhelmingly wide margin. Solve that and everything else fixes itself more or less instantly. Everything is now about money and extracting every single penny possible, instead of about actually achieving things. Even most 'entrepreneurs' are now just starting businesses primarily with the goal of selling them. Everything is broken, because of the pursuit of money became the goal, further compounding by everything being run by people who have no skills except the pursuit of money.
Money should be a means to achieve a thing, not the goal in and of itself. I think the most visible decline came with the increasingly overt goal to charge rent on friggin everything. That's simply not a sustainable or realistic economic model for society and consequently even if it might maximize corporate income in the short to mid-term, in the longterm it's equally catastrophic for them as well.
It’s clear to me that the smartest thing China ever did was to limit speculation in the markets. So many human capital in America is wasted pumping up valuations instead of actually building stuff
Every Jane Street hire could be building robots, but instead, they’re trading options and crypto and heck, even market making for prediction markets now
The positive aspect is that there is plenty if venture capital for innovators; the negative one is that those innovations are stifled by various extraction techniques that allow VCs and other investors to get a return on investment.
Crypto is a good example of how the equilibria is hard to maintain, and if the last cycle saw many interesting new products come to life, they all got crushed by ruthless profit-taking from early investors and team members.
> the smartest thing China ever did was to limit speculation in the markets. So many human capital in America is wasted pumping up valuations instead of actually building stuff
there was this Chinese company named Baofeng that built a stupid media player by "re-using" open source FFmpeg code, it managed to get itself publicly listed, then had its valuation went up like 50x for doing nothing other being accused for stealing FFmpeg code.
there were lots of discussions at that time how that happened and why the same level of speculation didn't happen on other public tech companies listed on Chinese market, the consensus was pretty sad - tech companies suitable for speculation are listed in the US by default, those listed on Chinese markets are 2nd tier or 3rd tier to start with, they don't offer any meaningful room for speculation.
Also hardwares availability. I saw some X threads that mentioned how US/EU robotics labs/companies need week to procure new hardwares, Chinese ones need days at most. Cant iterate quickly with that constraint.
Imagine you need weeks to start a new software module development and to procure cloud instances.
On one hand Boston Dynamics showed similar skill robot well before this demo, only without coreography, which is were most of the wow effect comes from here.
Things is american research is financed by outcome potential not for grandstanding, and free standing robot that can only do recorded coreography aren't that useful outside factory floors, and factory floors can use ceiling rails or wheels to better effect.
So yeah video is suler cool, but there isn't much to it beyond that to read in terms of capabilities. You seem just to be projecting what the truth you want to be on top of a funny dance.
China installed 295,000 industrial robots last year, more than the rest of the world combined, and has over 2M deployed total. China makes its own robots (57% indigenous) and its rate of robot deployment continues to grow year to year.
Meanwhile, the US installed 34,200, a decrease on the previous year, and virtually all of those were imported.
And what was the % of industrial capacity non automated over this percentage increase, because of course an industrializing country will have more manual processes to automate.
You asserted that American research is "financed by outcome potential not for grandstanding". But China is not just building dancing robots, they're also installing tons of industrial robots whose sole purpose is making money right now.
I visited China in November, the amount of different brands of electric vehicles is staggering. And even small hotels had robots delivering packages or food to the rooms.
What impressed me the most is the amount of EVs on the streets.
As European faced with similar pain points, I would assert it was having those MBAs offshoring everything with a colonial attittude, as if the nations on the received end would only take orders from their masters and not learn to master the technology themselves.
After a while, naturally the locals would buy the white label products that are anyway the same as the branded ones, many times produced on the same factory lines.
My father used to say, every company goes downhill when management takes over, meaning those straight out management schools without any actual business experience on what the company does, and he was kind of right, that is how we hand landed in late stage capitalism and entshitification, in the middle of geopolitics turn over.
These robots might not drive the car for us, but certainly will become part of some police containment unit, regardless if they are remote controlled or AI driven.
To best understand the speed of progress right now, take a look at the show from last year: https://www.youtube.com/watch?v=JIq_AM4q534
It looks like the difference between the Boston Dynamics robots 2016 vs 2021
The Spot dog (which inspired the Black Mirror "Metalhead" episode) in 2016 https://www.youtube.com/watch?v=tf7IEVTDjng
Atlas doing backflips in 2021 https://www.youtube.com/watch?v=FByY3tSx2Ak
So 5 years of progress within a year.
spot dog is hydraulically powered junk, unitree is motor driven from day one. Boston Dynamics was forced to switch to a motor driven architecture after it is proven by unitree.
Boston Dynamics is the follower here.
I look after one at a University that gets used for teaching & outreach & there's definitely no hydraulics on the thing.
why is hydraulics junk?
Perhaps because of the potentially slower actuation speed, but you also generally get a lot more power from hydraulics so im not sure one can claim it is junk. Far less acrobatic, but also far more sumo wrestler.
Some people on X are saying they're "just" cloning/copying "puppet" human movements.
I know very little about robotics, but given these appear totally free-standing, if that was the case (I personally don't think it is), wouldn't that imply they have the same centre of gravity and weight of limbs as humans? Surely they'd have to be able to balance themselves, and copying a human's movements "exactly" wouldn't work for their own motion otherwise?
I think when watching I saw one or two of the robots "judder" their feet a bit out of sync with others - this seems to imply they are capable of balancing their own motion a bit individually.
I've worked on much less expensive, much smaller humanoid robots.
These robots are certainly running through a scripted set of poses which has been extensively tested for the conditions (Humans would also be choreographed and have to hit certain marks at certain times). If you covered the stage in loose gravel or a thick carpet they'd all start falling over. The things the robots hold are almost certainly taped into their hands.
Despite that, this is a very impressive demo. Those robots are $40k+, they've got 20+ of them. And not a single one fell over. They're fast too - and there are a load of corners they could have cut, but they didn't.
The floor has two textures, it would have been easier without that. The humans right alongside them? Much less safety paperwork without them. The robot wearing trousers and a cape? Much easier without that. The fewer robots you have, the lower the chances on falls over landing their backflip. Lose the audience and record it in multiple takes. Hell, you could have human acrobats in robot costumes and it'd cost far less and be much easier.
So this demo is very much a costly signal of confidence.
> The things the robots hold are almost certainly taped into their hands.
You can clearly see that the robots change their grip of their sword, so it cannot be taped to their hands.
Probably a compliant magnetic coupling, as even simple force-sensing mechanical hands cost too much to practice back-flips. =3
Smaller platforms are actually harder to build: minimal power budget, weaker drive systems, less sensors, and fewer processing options.
Not a fan of bipedal platforms or 50kg of servos for a number of reasons.
Best regards. =3
> They're fast too
That was of of the two things that impressed me most, along with the choreography involving close and direct contact
It’s not a 1:1 human motion capture to servo translation. There is some work done to fix Center of gravity like you said and issues with friction and momentum.
The hard part with “autonomy” is interpretation of the environment and feeding that back into some control loop to accomplish a goal in real time. That is why most of these demos are basically recordings of movements, like choreography.
They're also interacting with the environment (vaulting boxes / walls), which implies they either know their 3D position very accurately, or they have some form of sensors and can adapt a bit.
As someone who owns a pair of Unitree G2s this blew my mind
what are you using those for ?
It looks like they have a point https://x.com/Osint613/status/2024068210660643009
Of course the robots have been pre-trained and the movements are scripted, and nobody is claiming otherwise. But there must be a lot of autonomous balancing taking place. At one point you can see the robots adjusting their feet slightly different although they are all in sync, and that catapult does not look like its movement is exactly the same every time. It is just super impressive.
Does anyone remember when Honda's Asimo robot clumsily fell down the stairs during a demonstration[1] and we thought we were safe from a robot invasion by just moving to the upper floor? That was about 20 years ago.
[1] https://www.youtube.com/watch?v=6mx6paHrnIE
The impressive part here isn't the movement itself. You can easily train a model to perform a "procedural animation" that includes a full body control policy. The hard part is making it reliable enough to perform long sequences of movements and adapting to differences in robot placement. In other words, performing a flawless stage play is the hardest part.
I'm afraid you might not understand what you're talking about. Animation is a geometry problem, while robotics is a dynamics problem. The latter is subject to constraints many times greater than the former. There is no such "easy" model as you imagined that can transform the former into the latter.
> balancing their own motion a bit individually.
check this 4 months old video below
https://www.youtube.com/watch?v=bPSLMX_V38E
I'd willing to bet that it is already close to impossible to get the robot lose its balance without some significant external forces.
Those are the same people that say that China is 30 years behind in chip manufacturing.
I don't know if you're aware but robots and chips are different things that require different expertise.
It's still a completely farcical attitude. At this point it's just a matter of how many years it'll take for chinese manufacturing to outstrip western expertise with chips, too. Ten years? Five? Two?
Very exciting times to live through.
Yes, the autonomy level of these robots was what I was yesterday emailing with my former colleagues we were wondering. Two months ago CNET & PC-Mag posted following video which suggests more about robots movements being assisted by humans. And it also shows Chinese have being edge of the development at that point.
https://www.youtube.com/watch?v=jXTibM33SDg
However, then another short video bit alike popped up and is puzzling too.
Apparently Unitree robot is playing pingpong match like a pro. Sorry about german announcer, I couldn't find with english.
https://www.youtube.com/shorts/BgD1ukTyNnw
There is another match viewable by pressing that "Robot plays ping ppng #robot" arrow.
How about that robot? Is it human assisted or not? Our opinions diverted, I'm quite sure it is assisted but my former colleague thinks it's got to be autonomous as it would be too difficult and slow to do that fast movements with remote control assisted robot.
It would be nice to hear opinions about that playing robot too if anyone could provide some insight in that.
edit: I think the serve waiting robot hand movement and after losing wiping left eye gesture as a disappointing a bit in my opinion gives up it's human. Or if not, why would a robot do such a human like gestures.
edit2: OK, good points, I see now. It's definitely a fake. Thanks to all who replied :)
that ping pong video is a CG robot, whether realtime superimposed or otherwise who knows. Look at the :27 when it gets out of tracking breaking all of physics, feet aren't planted to the ground, light, shadows.. etc.
here is a real video of a unitree robot playing ping pong https://www.youtube.com/watch?v=tOfPKW6D3gE
I think the ping pong match video might be misleading you. Based on the visual artifacts around the robot, the original footage likely had a human player that was swapped in with a robot in this video. It also has an altered content warning.
I can think that future use of pingpot robot is to replicate specific pro player style (from various recording) and be used to spar by pro players before their specific matches.
The ping pong video you linked is clearly fake. Look at the paddle... anyways...
Watch out, the two shorts you linked (both of robots playing ping-pong) are fake.
The pingpong video is very obviously computer generated. The robot feet give it away immediately
The second video you've linked is fake in every aspect in regards to the robot.
The robot is floating above the ground.
The paddle is phasing in and out of existence.
The robot has a realistic human hand and uses it to hit the ball.
The robot randomly turns around mid-air near the end of the video.
The robot looks nothing like a Unitree robot.
Oh, how could I forget, the entire robot looks so obviously fake even when disregarding all of the above that I can't believe you're even trying to analyze anything in that video.
I'm 99% sure that ping pong match is CGI. The whole robot has this green screen effect. Look at its feet. And at second 17 it just disappears entirely for a few frames.
I was born in 1989. The most impressive sudden technological advance I have experienced have been LLMs. This video is a good candidate for second place. I am mindblown... That they even dare having children dance with them. The trust they must place. An acquantance bought a chess board with a robot arm, and it accidentally broke his finger because he picked up a piece that the robot arm wanted to pick up. China isn't just a few hours in the future, more like decades it feels like.
Last time I had to implement a typed programming language, it was for short programs only. So for typing I just converted the program into a large set of constraints and throw z4 at it, asking to optimize for smallest possible types. If z4 could find types for every expressions in less than a few seconds then it was well typed.
You had to use a few trick for larger programs, but basically it managed to type any real programs and I never encountered an ambihuous case that caused a problem in practice. In case of failure, the small set of unsatisfiable constraints was easy to translate into nice error messages. This also allows for typing rules that are easy to state and can accomodate operators that adapts nicely to their environment.
I would understand if this approach would be frowned upon, but I still wonder if any serious language ever used this approach?
damn, now I would like to know what you were really answering to.
probably under a wrong thread, but can you point me to the one you intended? Really curious about that
whats z4
This is AMAZING.
We are definitely on an exponential in term of capabilities of humanoid robots. We are probably only years away from having a robot in the house, in construction of robots. Automating anything that a human can do is best done in a human sized robot.
But.
None of these are actually useful right now. I don't want something with the arm strength of a forklift taking care of my parents or kids. The demand for humanoid robots right now is like lift a fridge from a delivery truck to a house (aka more mobile forklift) or walk through toxic sewage to pull crates out. Super useful but basically just mobile cranes, which is a small market. China seems to be making the mistake of pushing a tech demo as a consumer product (we've all been on those projects...) which can make people hate the tech.
Build something people want, don't mandate what they want. We're like 3-4 generations from amazing, useful robots. I'll be scared when these things are minding a bunch of dogs on stage.
>> None of these are actually useful right now.
They could be really useful: without hesitation such humanoid could bring pack of explosives to the opposed treeline.
> I don't want something with the arm strength of a forklift taking care of my parents or kids.
A risk I never hear discussed is falling over and injuring children. Even the petite Unitree models are like a 70kg piece of furniture. Each year thousands get injured because of furniture falling over. I'd buy one immediately, but if I had kids or pets, I would wait for safety data on falls.
Humans have a similar weight and can also fall, so would you be satisfied with a fall rate the same or better than an average human?
Just submitted this as well. This is remarkable. Boston Dynamics has some catchup to do.
To be fair to BD, Atlas can lift 50 kg and a Unitree G1 can lift about 2 kg. An Atlas could literally pick up and throw a G1.
They are very different robots with very different goals, so it should be no surprise that the G1 appears much more agile.
Isn’t it easier to make an agile robot big than to make a big robot agile?
One big issue is the "joints". It's always a trade-off between mass, strength, speed, precision, and dexterity. State of the art is matching or exceeding organic joints on 2-3 axis, while being an order of magnitude below human/animal performance on the two other axis.
Human individual fingers can withstand internal loads of hundreds of newtons, (possibly a thousand for brief periods if you're a star rock-climber), while at the same being capable of tasks such as writing, which require high speeds and (sub-)millimeter precision, while also enjoying 4 degrees of freedom (5 for the thumb), and they're stuffed full of sensory organs to boot. Oh and they're also self-healing, so taking some damage during use is no problem and they will actually adapt to the task they're used for over time. Everything we can make compared to this is laughably primitive.
If you make an agile robot big, it now weighs more, which means its joints now need to handle greater loads, which makes it necessary to make them much bulkier and heavier.
A lot of this is just inherent limitations of electric motors and having to convert rotational energy. This gets heavy fast.
Decent synthetic electrically-driven muscle fibers would go a long way.
Yeah, when the G1s dance with kids, I realized how small they actually were. Definitely not the same category.
The H2 is what everyone is talking about now
some specs here: https://www.unitree.com/H2
claimed 3h battery life, can hold about 10% of its weight (7kg, with arms)
So Atlas can lift 7x the capacity. Even Digit, the tote-consolidating robot, can do 35lbs.
Unitree's demos are a lot of fun, and the antics of releasing the G1 to the public has certainly captured people's attention, but a "working" robot won't look, act, or develop from the G1 or even H2.
Unitree has plenty other industrial robots. https://www.unitree.com/ -> Click Robots
I wasn't trying to say that Unitree is somehow deficient. I'm sure they could build Atlas if they wanted.
My point was that BD could probably build a robot with the shown acrobatic capabilities, but they choose not to because their goal is to build robots that carry heavy loads for industrial applications.
I don't need a robot to lift more than 15lbs in order to do all my maid work
Focussing on load capacity is missing the forest for the trees
The point is that a robot with higher load capacity is necessarily less agile. BD's target market is industrial so their robots are necessarily larger and less agile.
The fact that Unitree's robots appear so acrobatic reflects that they are likely on par with BD in terms of capabilities but have a different target market.
Boston dynamics is far behind plus the robots are so cheap , even their dog is cheaper than BD. I dont think their humanoid can even catch up to this price. I am sure US Army and for the chinese counterpart Chinese army will be their biggest customers. But i wonder how will this workout in situations like Plane hijack, fire fighting and other such places where human lives cant be risked to save more human lives. (Please Dont downvote because your american patriotism is poked try replying.)
commenting on downvoting is against HN rules and typically comes with a compounding effect
I personally agree with the rest, recommend you remove that last bit about downvoting and sneering if you can still edit
What really stood out was that when they portrayed different important jobs it was all done by men and women were in the background as decoration/onlookers in awe. Very strange development.
I guess the US's answer is going to be gas-powered robots ;-).
Coal-powered steampunk robots would be awesome :D
Any information what battery life these things have? Would a human be able to outrun them given the need?
3 hours according to the website, and the battery looks to be quick swappable
Can it swap its own battery?
No, it cannot. That's why the right design for robots is for it to have two, so it doesn't have to do the Bemo fall.
but can you play games on its face like Bemo, Cozmo, and Vector?
I doubt so, but another one could in theory
I assume this is highly staged as a set routine?
No less impressive, but is it likely each robot autonomously learned a routine? Or just got programmed for a very exact act?
I'm not so sure what this question means... are you asking if China has AGI right now? I'm quite sure all similar performances done by humans are all staged.
Where human performances are staged, the human still needs be capable of learning the choreography. There's a difference between a robot learning the routine vs having it hardwired.
Perhaps improv theater is the better test for AGI
LLMs are already way too prepared for "Yes and..." improv, given GPT's ridiculous need to click-bait the end of every conversation.
Based on where other companies are right now, it's probably a pre programmed routine, but the robots autonomously balance themselves. Anything beyond that would be quite a large leap, and I think Unitree would've gloated about it way more outside of the gala. The robots' speed and consistency are still impressive, though.
The learning here is about what servor motors to control how much. But the performance was 100% staged and preprogrammed. Still impressive.
It is time to rewatch Terminator 2.
If the human-robot war is waged on a perfectly flat surface, we are fucked!
What is the most impressive is the robustness. Of course they are following a captured human routine, but they are facing so many disturbances from which they need to recover and keep following the desired trajectories, while under multiple constraints (movement ranges, not losing balance, etc).
You can see on the backflips that all robots landed quite differently, some with both knees on the ground, some with one, some with none. Yet all recovered gracefully and moved on to the next step of the choreography.
It is genuinely impressive, and scary.
Meanwhile in the west we are bickering like 10year olds.
Oh god the bickering. For self driving cars, lidar vs cameras is totally missing the point. Waymo can drive with cameras only. The AGI question is what decisions does it make when things go wrong.
Well, that is obviously another overcapacity issue, this time for robotics. In the last 10 weeks, there were -
Unitree's army of robots
https://www.youtube.com/watch?v=w4IOJH9Akhg
Robotera sword dancing robots
https://www.youtube.com/watch?v=9Ti9Mi8rbIQ
AGIbot's flying kicks
https://www.youtube.com/watch?v=QXnXdh6IEkA
LimX's Tron2 robot
https://www.youtube.com/watch?v=Ut3QFPr7hyo
one interesting observation is that none of those companies are located in Shenzhen, which arguably has the best supply chain for all electronics stuff. I guess those trillions $ spent on infrastructure paid off - Shenzhen didn't suck all talents into its proximity, it becomes an enabler for industries across the country.
Agibot is not in Shenzhen? I think what happened is that other major cities/provinces like Hangzhou, Suzhou, Shanghai or Hefei started bankrolling talents and enterprises so they too can have major high tech enterprises in their town
> Agibot is not in Shenzhen?
AGIBot is the poster boy of the municipal government of Shanghai.
> major cities/provinces like Hangzhou, Suzhou, Shanghai or Hefei started bankrolling talents and enterprises so they too can have major high tech enterprises in their town
State Capitalism at its best.
LimX looks to have a more humanoid offering now with a newer demo video than the one you linked
https://www.youtube.com/watch?v=0hIqs3TBb5g
Oh no, LinkedIn wall is leaking.
battery life measured in a handful of minutes too :-P
Tbf, if I did whay they do in this video, my battery life would more like a handful of seconds.
3 hours, so if 180m is a handful, but I don't agree, my micro-drone is a handful at 3-4m of flight
there is reason why in those 5 minutes you see them together in same shot with audience only 2 times and only for few seconds
there is reason why most of the shots are not wide angle showing whole scene, seems they learned their lesson from last year where you could easily see on the edges all the failures
this was heavily edited and repeated, I mean is it really surprising considering all CGI you see during whole gala? I watched whole 5 hours (though skimmed through a lot), they just can't make show same as seen by real people on the site, what you see in TV is very different from what audience has seen
edit: the whole gala show is recording, it would be impossible to organize such event across many cities with so many performances live, olympics opening ceremony is walk in the park compared to this
>this was heavily edited and repeated
While I don't know whether this was indeed broadcast live, at least this recording is missing a section since as the YT comments point out at 1:25 the staffs appear out of nowhere.
Can you elaborate on where? I'm not seeing anything at all at 1:25. This is a timestamp to 1:24: https://youtu.be/mUmlv814aJo?t=84
I was initially somewhat skeptical as well because this looks like a surprisingly massive leap in robotics capability, but haven't been able to find anything particularly sus.
Start watching at like 1:19 and see the kids don't have their own staff. Then at around 1:24 they do.
I think you're seeing a magic trick that's been around forever: https://www.youtube.com/watch?v=o-KDuOvMAfQ&t=80s
Ahhh! When he said staffs I assumed he meant people, not staves! Yeah, that's 100% CG. Here [1] you can see where the kids take a stance where it was to be edited in, and if you pause at the exact moment - it looks pretty bad, with something like a lightsaber just popping in the kid's hand.
Hmmm. Not sure what to make of this. I wish it was possible to see the raw unedited footage. It makes me question everything else.
[1] - https://youtu.be/mUmlv814aJo?t=82
If you watch at 0.25 speed, you can see the staff is actually a magician's cane. It's a coiled up bit of plastic that springs open when you release the ends. You can see similar tricks on YouTube, like https://www.youtube.com/watch?v=d-j6G4sPwWU
Have you seen the full gala? There are many sections where they interlace full intense cgi, it’s kinda like the new apple conference transitions where they purposefully don’t aim for reality, more like an enhanced experience that may or not have a purely live version.
Yeah, I noticed the heavy editing of the gala this year too and it was very disappointing. Even in tons of performances / dances where they really shouldn't have needed to there were obvious cinematic shots were not 'live' from the same recording. While for the robots I can imagine the pressure might have lead to editing a few takes, it really took away from the regular dancing performances and made it feel a bit more like watching a heavily edited music video.
1) Cool, but when are they actually going to drive my car for me?
2) Any semblance of American technological superiority is pure fantasy at this point. The only area where Americans are truly "advanced" is in selling overpriced SaaS products. There are dozens of Chinese startups with robots just like this—as seen at CES—yet Boston Dynamics is still treated like it’s some untouchable, DARPA-level tech.
3) A lot of this comes down to cost: you can either hire one American fresh grad or a Chinese PhD for the same price.
3) The second reason is cultural: Americans tend to buy solutions, while the Chinese prefer to build them. Even SMEs in China maintain internal dev teams to build custom software for the business, as opposed to paying Salesforce for what is essentially a glorified Excel sheet with sprinkles of automation.
4) America is facing its own innovator's dilemma. The country is currently being run by MBAs and salespeople focused on extracting every last dollar from the consumer instead of providing real value or innovating. Perhaps we're one step beyond the innovators dilemma. The innovators are dead and we are in the corporate greed stage.
5) Americans are completely oblivious to how advanced China has become because of the propaganda they're fed. My personal "aha" moment was when Chinese EVs hit my local market and completely obliterated legacy automakers on both features and price. The American "free" (lol) market is being guarded by politicians but that won't work for long.
#4 is the biggest problem, by an overwhelmingly wide margin. Solve that and everything else fixes itself more or less instantly. Everything is now about money and extracting every single penny possible, instead of about actually achieving things. Even most 'entrepreneurs' are now just starting businesses primarily with the goal of selling them. Everything is broken, because of the pursuit of money became the goal, further compounding by everything being run by people who have no skills except the pursuit of money.
Money should be a means to achieve a thing, not the goal in and of itself. I think the most visible decline came with the increasingly overt goal to charge rent on friggin everything. That's simply not a sustainable or realistic economic model for society and consequently even if it might maximize corporate income in the short to mid-term, in the longterm it's equally catastrophic for them as well.
The micro-financialization or wallstreet-ification of everything is a succinct way I try to describe it to normies
It’s clear to me that the smartest thing China ever did was to limit speculation in the markets. So many human capital in America is wasted pumping up valuations instead of actually building stuff
Every Jane Street hire could be building robots, but instead, they’re trading options and crypto and heck, even market making for prediction markets now
The positive aspect is that there is plenty if venture capital for innovators; the negative one is that those innovations are stifled by various extraction techniques that allow VCs and other investors to get a return on investment.
Crypto is a good example of how the equilibria is hard to maintain, and if the last cycle saw many interesting new products come to life, they all got crushed by ruthless profit-taking from early investors and team members.
> the smartest thing China ever did was to limit speculation in the markets. So many human capital in America is wasted pumping up valuations instead of actually building stuff
there was this Chinese company named Baofeng that built a stupid media player by "re-using" open source FFmpeg code, it managed to get itself publicly listed, then had its valuation went up like 50x for doing nothing other being accused for stealing FFmpeg code.
there were lots of discussions at that time how that happened and why the same level of speculation didn't happen on other public tech companies listed on Chinese market, the consensus was pretty sad - tech companies suitable for speculation are listed in the US by default, those listed on Chinese markets are 2nd tier or 3rd tier to start with, they don't offer any meaningful room for speculation.
Also hardwares availability. I saw some X threads that mentioned how US/EU robotics labs/companies need week to procure new hardwares, Chinese ones need days at most. Cant iterate quickly with that constraint.
Imagine you need weeks to start a new software module development and to procure cloud instances.
Unfair reading bordering with propaganda.
On one hand Boston Dynamics showed similar skill robot well before this demo, only without coreography, which is were most of the wow effect comes from here.
https://m.youtube.com/watch?v=UNorxwlZlFk
Heck check were they were 5 years ago
https://m.youtube.com/watch?v=fn3KWM1kuAw&pp=0gcJCUABo7VqN5t...
Things is american research is financed by outcome potential not for grandstanding, and free standing robot that can only do recorded coreography aren't that useful outside factory floors, and factory floors can use ceiling rails or wheels to better effect.
So yeah video is suler cool, but there isn't much to it beyond that to read in terms of capabilities. You seem just to be projecting what the truth you want to be on top of a funny dance.
China installed 295,000 industrial robots last year, more than the rest of the world combined, and has over 2M deployed total. China makes its own robots (57% indigenous) and its rate of robot deployment continues to grow year to year.
Meanwhile, the US installed 34,200, a decrease on the previous year, and virtually all of those were imported.
https://ifr.org/ifr-press-releases/news/global-robot-demand-...
Do these dance?
And what was the % of industrial capacity non automated over this percentage increase, because of course an industrializing country will have more manual processes to automate.
You asserted that American research is "financed by outcome potential not for grandstanding". But China is not just building dancing robots, they're also installing tons of industrial robots whose sole purpose is making money right now.
What is the automation gap tho
I visited China in November, the amount of different brands of electric vehicles is staggering. And even small hotels had robots delivering packages or food to the rooms.
What impressed me the most is the amount of EVs on the streets.
As European faced with similar pain points, I would assert it was having those MBAs offshoring everything with a colonial attittude, as if the nations on the received end would only take orders from their masters and not learn to master the technology themselves.
After a while, naturally the locals would buy the white label products that are anyway the same as the branded ones, many times produced on the same factory lines.
My father used to say, every company goes downhill when management takes over, meaning those straight out management schools without any actual business experience on what the company does, and he was kind of right, that is how we hand landed in late stage capitalism and entshitification, in the middle of geopolitics turn over.
These robots might not drive the car for us, but certainly will become part of some police containment unit, regardless if they are remote controlled or AI driven.
Looks fake.