Also, you should probably give some thought to inventing a practical fucking reason for humanoid robots before you launch. When it comes to running 100m footraces, I'm well-satisfied with the...
Still, experts say humanoid robots are still mostly used for demonstrations, performances and research — at least for now — and it will still take time to achieve mass real-world deployment.
Also, you should probably give some thought to inventing a practical fucking reason for humanoid robots before you launch. When it comes to running 100m footraces, I'm well-satisfied with the service of my current provider. Thanks.
The point is not the activity itself, but the fact that it can mimic complex movements, like running, and be very good at it. If you compare this to what humanoid robots could do and how advanced...
The point is not the activity itself, but the fact that it can mimic complex movements, like running, and be very good at it. If you compare this to what humanoid robots could do and how advanced robotics like this was 10 years ago, the improvement is super impressive.
Assuming they keep improving, this could have incredible uses. Think firefighters who can walk into burning buildings no problem. The potential is very exciting. On a more sombre note, humans will human and somebody is going to stick a gun on these things as soon as possible.
This is the real answer. It’s not designed to replace Olympians, it’s a demo to show how capable it is with something. Next they’ll show it threading the eyes of ten thousand needles or something...
This is the real answer. It’s not designed to replace Olympians, it’s a demo to show how capable it is with something. Next they’ll show it threading the eyes of ten thousand needles or something to highlight it’s dexterity.
Throwing ten thousand tiny balls into ten thousand slots against a crowd of 1.5-2.0m tall dummies, more like. Gotta demonstrate useful dexterity the ultra rich can see needing. Then they can buy...
Throwing ten thousand tiny balls into ten thousand slots against a crowd of 1.5-2.0m tall dummies, more like. Gotta demonstrate useful dexterity the ultra rich can see needing. Then they can buy cheapened human labour for the ten thousand needles.
The use case as I understand it is pretty much “drop into the billions of situations that are already designed around humans doing the thing, and have the robot do the thing without needing to...
The use case as I understand it is pretty much “drop into the billions of situations that are already designed around humans doing the thing, and have the robot do the thing without needing to retrofit the whole environment for them”. That and it’s goddamn cool, of course.
So as with a lot of tech recently: a kind of amazing leap that would’ve been sci-fi not so long ago, and one that takes the complexity of the human environment as it is (physical now, as well as linguistic and visual over the last few years) rather than fighting a losing battle to impose computational order on chaotic reality. But all the wonder is going to be sucked out of it because we all know it’ll just be used to siphon yet more money up towards like six dudes, despite its huge potential to do actual good otherwise.
I think the approach is pretty flawed, because humans and machines are put together in vastly different ways. The human body has over 600 different muscles. That's 600 actuators that can...
I think the approach is pretty flawed, because humans and machines are put together in vastly different ways.
The human body has over 600 different muscles. That's 600 actuators that can independently control motion to different parts of the body.
Replicating that with a robot is ridiculously expensive, and will always remain ridiculously expensive. Even if you're talking about a $20 servo, that's $12,000 in servos alone. We don't have a way to grow robots either, they need to be meticulously assembled, ironically, by hand in most cases, so at the end of the day, building one with the full capabilities of a human is astronomically expensive.
At the same time, humans aren't even that good at most physical tasks. When we mow a lawn, we use a device with a giant handle and a gas engine that we use our dumb feet to push around. A robot lawnmower does the same thing with a handful of motors and battery, and it does it way, way cheaper than a humanoid could.
So if you want a robot around your house to help you with a few tasks, say, vacuuming, mowing your lawn, and folding your laundry, are most people going to buy the $100,000 humanoid that will be equally expensive to fix when it breaks, or the $200 robot vacuum, the $400 robot lawnmower, and the $1000 robot laundry folding machine, all of which can work at the same time?
Id guess there probably are places that a humanoid robot makes sense (I actually struggle to think of one right now though), inside people's homes probably isn't that place though.
Once you can grow them in vats or replicators and just recycle them when they break, maybe, but beyond that I don't see the financials ever making sense.
I think you’re right about any research aiming for absolute 1:1 human parity, but I do think there will be a genuine market for broadly human shaped and sized robots, with bipedal motion to...
I think you’re right about any research aiming for absolute 1:1 human parity, but I do think there will be a genuine market for broadly human shaped and sized robots, with bipedal motion to traverse any human-navigable terrain, a 1m-ish reach on each of two actuators, a roughly similar range of gross motions, a main camera around 170cm from the ground with the associated field of view, etc etc
For what it’s worth I wasn’t even entertaining home use as a possibility, and I was thinking ballpark $25-50k cost. That’s where companies are going to start having them operate existing machines because it’s not even worth the downtime to retrofit those machines with automated controls, let alone the engineering time to design the custom modifications they’d need. Same goes for working in bits of the power plant crawlspace that were sized and shaped with very specific constraints in mind, or carrying equipment across the walkways with all the steps that bounce a little too much and have very specific weight tolerances, or reaching the lever while standing on the foot pedal, or whatever.
Could a skilled engineer design something custom for each of those jobs? Absolutely! Will they even get to the end of the meeting about timelines before the director has seen that engineer time cost > robot cost and bought one of these instead? I doubt it. And that goes triple once companies have a few around from previous projects and know they can repurpose them for new projects without buying again or redesigning.
Task specific robots make sense for home users who need mass produced minimum cost, and for new commercial projects where they can build for automated efficiency from the start. General purpose ones that more or less fit the human mould make sense where you want to automate, but retrofitting your existing warehouse, or fleet of excavators, or printing press, or whatever would cost mid six figures or more.
I think a humanoid form factor makes the most in areas where robots will be working alongside humans: medical care (being able to lift a person or change their diaper) and trades (where being able...
I think a humanoid form factor makes the most in areas where robots will be working alongside humans: medical care (being able to lift a person or change their diaper) and trades (where being able to squeeze into a tight space without any attendant back problems would be useful).
Even if they have fewer applications than anticipated, equipment developed for humanoid robots will make its away into other form factors.
You may be right, and I'm guessing that's what all these companies tied to humanoid robots are hedging for, but I've had a really hard time thinking of even an industrial use case where humanoids...
You may be right, and I'm guessing that's what all these companies tied to humanoid robots are hedging for, but I've had a really hard time thinking of even an industrial use case where humanoids would be better suited to a wheeled, tracked, or quadruped robot.
The human shape is mostly an evolutionary compromise based on using existing limbs and adapting two of them for tool use and fine motor manipulation instead of locomotion. It's really not beneficial for much except for being an exact stand-in for a human, like for crash testing, or testing chairs or something like that.
I think it's the second part much more than the first to be honest. This kind of dexterity and motor control is impressive and genuinely useful, the humanoid form factor probably isn't though. Its...
The use case as I understand it is pretty much “drop into the billions of situations that are already designed around humans doing the thing, and have the robot do the thing without needing to retrofit the whole environment for them”. That and it’s goddamn cool, of course.
I think it's the second part much more than the first to be honest. This kind of dexterity and motor control is impressive and genuinely useful, the humanoid form factor probably isn't though.
Its a lot easier and cheaper to design robots that can navigate specific environments to do specific tasks that they can do a lot better than a humanoid can versus trying to design one that can replicate everything a human can do. Still, if you want to build a laundry folding robot, or a firefighting robot, it needs to be able to have human level dexterity and object awareness, so the capabilities demoed at events like this are important.
To be fair to China (something I don't often do), America has been running DARPA challenges forever and their entire funding and stated purpose is military. I would just assume that whatever...
To be fair to China (something I don't often do), America has been running DARPA challenges forever and their entire funding and stated purpose is military. I would just assume that whatever Chinese researchers show off on this front is part of their military arsenal parade.
They're not worried about replacing human labour, but they sure can manufacture and sell everyone else machines to replace human labour, AND machines to mow down protesters against replacement of human labour.
I dislike the emphasis on them breaking human records in the headline and article, as if pushing the idea robots are taking over. Of course a robot can go faster than a man. So can a car. What's...
I dislike the emphasis on them breaking human records in the headline and article, as if pushing the idea robots are taking over. Of course a robot can go faster than a man. So can a car. What's new?
I don't have much to say about the event itself. I complain about AI a lot but I'm fairly neutral about humanoid robots ; I think it's very possible they'll be a net positive for the world, but the current capitalist system we live in makes it hard to imagine that a major technological breakthrough will ever lead common people to easier lives instead of making economy move faster and thus removing jobs and making the ones that remain harder.
Being on two legs is very different than being on wheels. It’s not that robots are taking over, it’s that a robot being able to run on two legs around a curved track at remotely human comparable...
Being on two legs is very different than being on wheels.
It’s not that robots are taking over, it’s that a robot being able to run on two legs around a curved track at remotely human comparable speeds would be unthinkable 10 years ago.
Although the differences in hardware make it difficult to compare in any real sense. For instance, I’d like to know more about the design limitation in terms of elastic recoil, range of movement...
Although the differences in hardware make it difficult to compare in any real sense. For instance, I’d like to know more about the design limitation in terms of elastic recoil, range of movement of the joints, height etc. In humans, these are limited by the genetic material you get, but in a mechanic system are designed to choices which can be optimised.
I remember at a previous Paralympics there was concern about an advantage over others due to too much elastic recoil (although from what I’ve read I’m not sure that has actually shown to be beneficial). My point being that when you can design a mechanical system from scratch it’s a very different ball game.
On a similar note, it would be interesting to see what the maximum speed you could get a human sprinter with some non wheel based technology (like those bionic boots or similar).
Don’t get me wrong, it is very impressive, but I do find the subtext with robots is “they are faster, smarter and better than you. Prepare to be replaced, and kowtow to your technology overlords” which I find rather distasteful.
You don’t need to compare it against humans. It’s better to compare it against robots 10 years ago. This just wasn’t possible then. Walking is hard. On a signal controls level, it’s so far away...
Although the differences in hardware make it difficult to compare in any real sense.
You don’t need to compare it against humans. It’s better to compare it against robots 10 years ago. This just wasn’t possible then. Walking is hard. On a signal controls level, it’s so far away from a 4 wheel vehicle in complexity it’s like comparing an ant to an elephant. Running is even harder. Trying to calculate how many force the motors in each joint needs to counterbalance the normal forces when landing is an immensely difficult task. It’s practically impossible in an analytic sense, hence why we needed more complex numerical systems estimators (eg, a neural network most likely).
Why compare it to humans? The net of it is that robots are getting competent at managing incredibly complex physics signal problems, something that previously only humans could do. How they do it is less relevant than the fact that they can do these types of loosely constrained problems.
I know the article mentions “smashing human records” but it’s just an arbitrary benchmark. It’s like saying “X sold more than 100,000 copies”. 100,000 is an arbitrary number that looks good because we happen to use base 10.
“they are faster, smarter and better than you. Prepare to be replaced, and kowtow to your technology overlords”
Are you sure that the subtext is there and it’s not projection? For what it’s worth, I don’t see that subtext, especially because it’s a Chinese piece. From talking with people I know in China, robotics and AI are seen generally very positively, and the news would be taken in a positive light, not a “you will be jobless flesh puppet” light. With the population pyramid being what it is in China, increasing per capita efficiency is the key to continued economic growth, and automation is seen as a key component of that.
That's literally the whole point of the article though. If it was about just achieving an arbitrary goal, say, 10x faster than a robot from a decade ago, it'd say that, but it's specifically about...
You don’t need to compare it against humans. It’s better to compare it against robots 10 years ago.
That's literally the whole point of the article though. If it was about just achieving an arbitrary goal, say, 10x faster than a robot from a decade ago, it'd say that, but it's specifically about beating human records. The subtext there is pretty clear.
I’m aware it’s a big improvement in robotics terms, hence why I specifically said it’s very impressive. As for comparing to humans and using 100m as an arbitrary bench mark, it is not. The robots...
I’m aware it’s a big improvement in robotics terms, hence why I specifically said it’s very impressive. As for comparing to humans and using 100m as an arbitrary bench mark, it is not. The robots are built in a human shape and directly compared to the human speed benchmark, as such it is an intentional direct comparison. Surely a more useful benchmark for robotics would be least falls per 100m or most efficient energy use? Doing it on an athletics track just reinforces the direct comparison to me.
As for your second point, your right there is probably some partial projection at least. Several news articles that mention this do frame it this way, and most robotic advances directly compared to humans endeavours are framed this way. I do not know enough about the Chinese zeitgeist to comment on how these things are viewed in China, but I remain skeptical enough of advances which are inherently centralised improving the power of individual workers in a capitalistic economy- but I’ll leave it there to avoid going too far off topic.
But this is not a case of "of course" at all. This is a new world record, and technological progress that just wasn't possible a few years ago. Basically a breakthrough success in engineering,...
Of course a robot can go faster than a man.
But this is not a case of "of course" at all. This is a new world record, and technological progress that just wasn't possible a few years ago. Basically a breakthrough success in engineering, which is why this is news.
No, I understand that, don't get me wrong. It's the way the headline is presented in a 'robots are taking over!' manner that bugs me. Perhaps I'm overdoing the media skepticism (not the right word...
No, I understand that, don't get me wrong. It's the way the headline is presented in a 'robots are taking over!' manner that bugs me. Perhaps I'm overdoing the media skepticism (not the right word but it's not coming) but I feel too much is better than too little.
It feels like chickens cheering for the farm getting a pie machine. (Chicken Run, YouTube, 0:23) Are apple pies terrible for chickens? Of course not, they'd absolutely love it. But that's not what...
It feels like chickens cheering for the farm getting a pie machine. (Chicken Run, YouTube, 0:23) Are apple pies terrible for chickens? Of course not, they'd absolutely love it. But that's not what the machine is for.
Chinese humanoid robots broke records set by humans, including beating Usain Bolt’s 100-meter sprint world record, on the opening day of the Olympics-like World Humanoid Robot Games in Beijing on Saturday.
The five-day games, now in its second year, are a spectacle demonstrating China’s rapid progress in advanced robotics as the technology race with the U.S. heats up, with 51 events and more than 1,000 competitions taking place including running, table tennis and soccer.
At a 100-meter sprint, a humanoid robot achieved a result of 9.39 seconds, beating the human record of 9.58 seconds set by Jamaican athlete Bolt in 2009.
In a standing high jump, a humanoid robot was able to reach 2.88 meters, well above the 0.95 meters best result by a humanoid in last year’s first edition of the games. It surpassed the human high jump record of 2.45 meters set by Cuba’s Javier Sotomayor in 1993.
Also, you should probably give some thought to inventing a practical fucking reason for humanoid robots before you launch. When it comes to running 100m footraces, I'm well-satisfied with the service of my current provider. Thanks.
The point is not the activity itself, but the fact that it can mimic complex movements, like running, and be very good at it. If you compare this to what humanoid robots could do and how advanced robotics like this was 10 years ago, the improvement is super impressive.
Assuming they keep improving, this could have incredible uses. Think firefighters who can walk into burning buildings no problem. The potential is very exciting. On a more sombre note, humans will human and somebody is going to stick a gun on these things as soon as possible.
This is the real answer. It’s not designed to replace Olympians, it’s a demo to show how capable it is with something. Next they’ll show it threading the eyes of ten thousand needles or something to highlight it’s dexterity.
Throwing ten thousand tiny balls into ten thousand slots against a crowd of 1.5-2.0m tall dummies, more like. Gotta demonstrate useful dexterity the ultra rich can see needing. Then they can buy cheapened human labour for the ten thousand needles.
The use case as I understand it is pretty much “drop into the billions of situations that are already designed around humans doing the thing, and have the robot do the thing without needing to retrofit the whole environment for them”. That and it’s goddamn cool, of course.
So as with a lot of tech recently: a kind of amazing leap that would’ve been sci-fi not so long ago, and one that takes the complexity of the human environment as it is (physical now, as well as linguistic and visual over the last few years) rather than fighting a losing battle to impose computational order on chaotic reality. But all the wonder is going to be sucked out of it because we all know it’ll just be used to siphon yet more money up towards like six dudes, despite its huge potential to do actual good otherwise.
I think the approach is pretty flawed, because humans and machines are put together in vastly different ways.
The human body has over 600 different muscles. That's 600 actuators that can independently control motion to different parts of the body.
Replicating that with a robot is ridiculously expensive, and will always remain ridiculously expensive. Even if you're talking about a $20 servo, that's $12,000 in servos alone. We don't have a way to grow robots either, they need to be meticulously assembled, ironically, by hand in most cases, so at the end of the day, building one with the full capabilities of a human is astronomically expensive.
At the same time, humans aren't even that good at most physical tasks. When we mow a lawn, we use a device with a giant handle and a gas engine that we use our dumb feet to push around. A robot lawnmower does the same thing with a handful of motors and battery, and it does it way, way cheaper than a humanoid could.
So if you want a robot around your house to help you with a few tasks, say, vacuuming, mowing your lawn, and folding your laundry, are most people going to buy the $100,000 humanoid that will be equally expensive to fix when it breaks, or the $200 robot vacuum, the $400 robot lawnmower, and the $1000 robot laundry folding machine, all of which can work at the same time?
Id guess there probably are places that a humanoid robot makes sense (I actually struggle to think of one right now though), inside people's homes probably isn't that place though.
Once you can grow them in vats or replicators and just recycle them when they break, maybe, but beyond that I don't see the financials ever making sense.
I think you’re right about any research aiming for absolute 1:1 human parity, but I do think there will be a genuine market for broadly human shaped and sized robots, with bipedal motion to traverse any human-navigable terrain, a 1m-ish reach on each of two actuators, a roughly similar range of gross motions, a main camera around 170cm from the ground with the associated field of view, etc etc
For what it’s worth I wasn’t even entertaining home use as a possibility, and I was thinking ballpark $25-50k cost. That’s where companies are going to start having them operate existing machines because it’s not even worth the downtime to retrofit those machines with automated controls, let alone the engineering time to design the custom modifications they’d need. Same goes for working in bits of the power plant crawlspace that were sized and shaped with very specific constraints in mind, or carrying equipment across the walkways with all the steps that bounce a little too much and have very specific weight tolerances, or reaching the lever while standing on the foot pedal, or whatever.
Could a skilled engineer design something custom for each of those jobs? Absolutely! Will they even get to the end of the meeting about timelines before the director has seen that engineer time cost > robot cost and bought one of these instead? I doubt it. And that goes triple once companies have a few around from previous projects and know they can repurpose them for new projects without buying again or redesigning.
Task specific robots make sense for home users who need mass produced minimum cost, and for new commercial projects where they can build for automated efficiency from the start. General purpose ones that more or less fit the human mould make sense where you want to automate, but retrofitting your existing warehouse, or fleet of excavators, or printing press, or whatever would cost mid six figures or more.
I think a humanoid form factor makes the most in areas where robots will be working alongside humans: medical care (being able to lift a person or change their diaper) and trades (where being able to squeeze into a tight space without any attendant back problems would be useful).
Even if they have fewer applications than anticipated, equipment developed for humanoid robots will make its away into other form factors.
You may be right, and I'm guessing that's what all these companies tied to humanoid robots are hedging for, but I've had a really hard time thinking of even an industrial use case where humanoids would be better suited to a wheeled, tracked, or quadruped robot.
The human shape is mostly an evolutionary compromise based on using existing limbs and adapting two of them for tool use and fine motor manipulation instead of locomotion. It's really not beneficial for much except for being an exact stand-in for a human, like for crash testing, or testing chairs or something like that.
I think it's the second part much more than the first to be honest. This kind of dexterity and motor control is impressive and genuinely useful, the humanoid form factor probably isn't though.
Its a lot easier and cheaper to design robots that can navigate specific environments to do specific tasks that they can do a lot better than a humanoid can versus trying to design one that can replicate everything a human can do. Still, if you want to build a laundry folding robot, or a firefighting robot, it needs to be able to have human level dexterity and object awareness, so the capabilities demoed at events like this are important.
To be fair to China (something I don't often do), America has been running DARPA challenges forever and their entire funding and stated purpose is military. I would just assume that whatever Chinese researchers show off on this front is part of their military arsenal parade.
They're not worried about replacing human labour, but they sure can manufacture and sell everyone else machines to replace human labour, AND machines to mow down protesters against replacement of human labour.
The practical reason is pretty obvious - automate human labor.
and soldiers to protect the owner class from the displaced human labor
I dislike the emphasis on them breaking human records in the headline and article, as if pushing the idea robots are taking over. Of course a robot can go faster than a man. So can a car. What's new?
I don't have much to say about the event itself. I complain about AI a lot but I'm fairly neutral about humanoid robots ; I think it's very possible they'll be a net positive for the world, but the current capitalist system we live in makes it hard to imagine that a major technological breakthrough will ever lead common people to easier lives instead of making economy move faster and thus removing jobs and making the ones that remain harder.
Being on two legs is very different than being on wheels.
It’s not that robots are taking over, it’s that a robot being able to run on two legs around a curved track at remotely human comparable speeds would be unthinkable 10 years ago.
Although the differences in hardware make it difficult to compare in any real sense. For instance, I’d like to know more about the design limitation in terms of elastic recoil, range of movement of the joints, height etc. In humans, these are limited by the genetic material you get, but in a mechanic system are designed to choices which can be optimised.
I remember at a previous Paralympics there was concern about an advantage over others due to too much elastic recoil (although from what I’ve read I’m not sure that has actually shown to be beneficial). My point being that when you can design a mechanical system from scratch it’s a very different ball game.
On a similar note, it would be interesting to see what the maximum speed you could get a human sprinter with some non wheel based technology (like those bionic boots or similar).
Don’t get me wrong, it is very impressive, but I do find the subtext with robots is “they are faster, smarter and better than you. Prepare to be replaced, and kowtow to your technology overlords” which I find rather distasteful.
You don’t need to compare it against humans. It’s better to compare it against robots 10 years ago. This just wasn’t possible then. Walking is hard. On a signal controls level, it’s so far away from a 4 wheel vehicle in complexity it’s like comparing an ant to an elephant. Running is even harder. Trying to calculate how many force the motors in each joint needs to counterbalance the normal forces when landing is an immensely difficult task. It’s practically impossible in an analytic sense, hence why we needed more complex numerical systems estimators (eg, a neural network most likely).
Why compare it to humans? The net of it is that robots are getting competent at managing incredibly complex physics signal problems, something that previously only humans could do. How they do it is less relevant than the fact that they can do these types of loosely constrained problems.
I know the article mentions “smashing human records” but it’s just an arbitrary benchmark. It’s like saying “X sold more than 100,000 copies”. 100,000 is an arbitrary number that looks good because we happen to use base 10.
Are you sure that the subtext is there and it’s not projection? For what it’s worth, I don’t see that subtext, especially because it’s a Chinese piece. From talking with people I know in China, robotics and AI are seen generally very positively, and the news would be taken in a positive light, not a “you will be jobless flesh puppet” light. With the population pyramid being what it is in China, increasing per capita efficiency is the key to continued economic growth, and automation is seen as a key component of that.
That's literally the whole point of the article though. If it was about just achieving an arbitrary goal, say, 10x faster than a robot from a decade ago, it'd say that, but it's specifically about beating human records. The subtext there is pretty clear.
I’m aware it’s a big improvement in robotics terms, hence why I specifically said it’s very impressive. As for comparing to humans and using 100m as an arbitrary bench mark, it is not. The robots are built in a human shape and directly compared to the human speed benchmark, as such it is an intentional direct comparison. Surely a more useful benchmark for robotics would be least falls per 100m or most efficient energy use? Doing it on an athletics track just reinforces the direct comparison to me.
As for your second point, your right there is probably some partial projection at least. Several news articles that mention this do frame it this way, and most robotic advances directly compared to humans endeavours are framed this way. I do not know enough about the Chinese zeitgeist to comment on how these things are viewed in China, but I remain skeptical enough of advances which are inherently centralised improving the power of individual workers in a capitalistic economy- but I’ll leave it there to avoid going too far off topic.
See what I replied to user shu!
But this is not a case of "of course" at all. This is a new world record, and technological progress that just wasn't possible a few years ago. Basically a breakthrough success in engineering, which is why this is news.
No, I understand that, don't get me wrong. It's the way the headline is presented in a 'robots are taking over!' manner that bugs me. Perhaps I'm overdoing the media skepticism (not the right word but it's not coming) but I feel too much is better than too little.
It feels like chickens cheering for the farm getting a pie machine. (Chicken Run, YouTube, 0:23) Are apple pies terrible for chickens? Of course not, they'd absolutely love it. But that's not what the machine is for.
Clanker Olympics