Martial arts robots at 2026 Spring Festival Gala [video]
youtube.com
youtube.com
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.
Boston Dynamics is the follower here.
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.
Very exciting times to live through.
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.
My point is that if you interrupt the sequence by pushing it over or standing in front of it, the unit will not be able to recover or mitigate the issue. I’m not sure about the models in the video but they should enter a safe state, hopefully gracefully.
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 :)
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.
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.
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.
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.
That was of of the two things that impressed me most, along with the choreography involving close and direct contact
You can clearly see that the robots change their grip of their sword, so it cannot be taped to their hands.
With the poles at the 1 minute mark, the robots enter holding them and their left hand never moves on the pole. Also note the stationary hand is matte grey while the moving hand is metallic silver.
Likewise with the wine gourds (?) at 2m30s and the nunchucks at 3m40s.
It’s a completely sensible design decision, much simpler to do cartwheels and vaults if you don’t have super delicate fingers fitted.
Not a fan of bipedal platforms or 50kg of servos for a number of reasons.
Best regards. =3
An Aldebaran Nao can fall over with no damage because it’s only 5kg and 58cm. And you can use relatively low power motors, so nobody can lose a finger to crushing in the joints.
But you miss out on the benefits of being able to operate in a human centric world - you’ll never get a Nao to climb stairs, open a door, or carry a cup of coffee.
They were overpriced, hard to repair, and the university lab would lose one every few weeks. The consumer grade internal ribbon interconnect harness would usually tear eventually.
Those 3kW to 5kW brush-less servos in the biped dance video are often only 12:1 planetary reduction, and are almost certainly backdrivable. As long as the platform has a safe-fall canned pose, than damage should be limited to shear pins in the worst case falls.
Fun to see people pushing the design limits. =3
The motion is certainly scripted, but the exact mechanics in play there almost certainly vary radically from take to take. Imagine something simple like a pool/billiards break. Even if you set up a machine to rack the balls and break them in as close to identical as possible, you'd get wildly different results each time. And the dynamics in this motion is going to dwarf that.
This would also be normal for human performers - touring ballet companies travel with their own flooring the dancers are used to pirouetting on.
At the 40 second mark every robot does a backflip then when landing hops their supporting leg while pointing the toe of their working leg. Which works fine and looks great! But they arrive in that pose with a certain amount of momentum and needing a certain amount of grip on the floor.
So this is a rehearsed, tested performance - not proof we’ll have firefighter robots doing parkour through burning buildings any time soon!
I wonder how similar friction is across varieties of standard issue corporate carpet...
Human acrobats do the same, they know a few fixed tricks and practice them often in order to stay in shape. The point here is to compare the dynamic abilities of bots vs those of humans, not vs some pink unicorns - the latter comparison isn't informative because there aren't any jobs for pink unicorns just yet.
> So this is a rehearsed, tested performance - not proof we’ll have firefighter robots doing parkour through burning buildings any time soon!
A bot assisted by humans can do better than them in a burning building, the lack of need for heavy SCBAs would allow the bot to perform acrobatics scripted by a human at the time of action.
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.
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
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.
Again, see crypto as a prime example - because, at one point, you could command a valuation that was simply not tethered (heh) to reality, you had all these now-dead L1 chains raising $200M+ at $3-5B valuations.
This also leads to a situation where you only end up funding digital plays because the metrics there can be anything. You had these crypto companies raise based on "growth" when that growth was simply coins produced out of thin air and wallets created by the millions with a script.
You can't do that if you're building actual physical products
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.
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.
Once China built an industrial base, and has cheap energy sources, you cannot directly out compete it. You can only try to maintain your own industrial base by locking competitors out of your market. There is no other way. In the end that's the result of globalization - US&EU companies thought they can produce cheap, sell expensive. Instead they trained their own competition, and due to weak IP laws enabled this exodus of industry know-how to China.
Now, if China was a smaller country - let's say Japan or Turkey - this wouldn't be such a huge problem. But for the global economy, having a single country that produces 80% of all consumer goods is also a huge problem. That was never the case before with the US (except maybe directly after WW2). US+Europe, Japan+Korea, Canada, Australia supply chain was much more diverse and distributed.
What happens now is dangerous because in the end the profits are not spread across the world, and economies of scale cause this monopoly to appear, which will be hard to mitigate.
Can countries "slow" down China and move production to diversified locations? For that to work, coordinated tariffs for advanced goods from China would have to be introduced, and production reallocated to multiple other countries - very difficult to execute..
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.
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-...
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.
Ah yes that explains Tesla...
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.
What impressed me the most is the amount of EVs on the streets.
Imagine you need weeks to start a new software module development and to procure cloud instances.
They'd only have been 10 in 1999 when cell phones were pretty ubiquitous among adults. I'd say they were basically past it already.
I was born ten years earlier and I'd agree that the internet is likely the biggest change in my lifespan, but I'm not sure I'd say it was as sudden as the past 2 years of AI.
If I look at the most useful technologies around me, then Google Maps ranks high. But it wasn't sudden. It was on the desktop first. And then slowly crept through mobile.
LLMs on the other hand, suddenly just kind of appeared.
They are very different robots with very different goals, so it should be no surprise that the G1 appears much more agile.
some specs here: https://www.unitree.com/H2
claimed 3h battery life, can hold about 10% of its weight (7kg, with arms)
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.
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.
Focussing on load capacity is missing the forest for the trees
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.
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.
I personally agree with the rest, recommend you remove that last bit about downvoting and sneering if you can still edit
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.
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 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.
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.
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.
Anyways these robots are cool and I hope we send robots to Mars instead of human astronauts.
They could be really useful: without hesitation such humanoid could bring pack of explosives to the opposed treeline.
Also, current tech could be useful as a shopping assistant, to carry the groceries for people who can't, for one reason or another. Though the other post about tipping safety does have a point.
Robots in such an environment are designed with the appropriate affordances so that they cannot use too much force... but the concern about weight I suppose is quite salient.
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?
Was intended for https://news.ycombinator.com/item?id=47025885
And z4 is, of course z3 (no I don't have special access to next-gen version of z3)
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.
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.
Reason: otherwise, you will be asking what happens when the shot is switched between the stage and the audience.
Now the nice thing about demographics is that it's eminently predictable. IMO this is why China is going all in on robotics.
No less impressive, but is it likely each robot autonomously learned a routine? Or just got programmed for a very exact act?