In spite of the demonstration videos of robots running and jumping and carrying boxes and slowly sorting packages, none of them are anywhere close to being ready to handle construction jobs, or most manufacturing jobs. And, they can’t design or make the machines that automate the processes.
It’s a billionaire’s dream that this is happening now, it’s not. One of the major hurdles is power consumption, as we’re all seeing while they build out these “data centers”. Robots and computers cannot match the efficiency and self-sufficiency of a human. In a picture:
Yep. I work in prosthetics, people tend to overestimate current battery technology and underestimate how much energy it takes to move around a couple hundred pounds.
Even with batteries that cost a small fortune, we haven’t been able to create a lower limb powered prosthetic that can keep up with the human body for more than a couple hours.
There’s a lot of hype about the humanoid robotics coming out of China, but most everything is just marketing science fiction. Once you factor in the initial cost, power consumption for processing data, multiple batteries per unit, and the massive amount of general maintenance… It becomes pretty apparent that things aren’t going to pan out anytime soon.
I build limbs out of reinforced carbon fiber and titanium and even those acquire a ton of wear in a relatively small amount of time. Any bipedal robot with plastic parts, made to do intensive labour is going to start falling apart almost immediately. The human body takes a ton of damage throughout the day as well, but it’s self repairing.
The equation changes a lot when you start altering the assumptions.
Who says the robots have to have batteries? Human workers drag power cables all over construction sites. Why can’t a robot be built that’s smart enough to manage its own power cord? Power cord management seems simple compared to doing meaningful construction work.
The wear and tear is a legitimate concern, but fixed industrial robots have been operating for many decades. The wear issues are obviously solvable in the right context.
Keep in mind, prosthetics may not be the perfect analogy. We imagine these things as humanoid, but they need not be. You might have a drywall hanging robot that has suction cups to lift up panels and a screw gun to affix them. They might not even have legs. A welding robot could just be a welding tool on the end of a long boom arm.
There’s really no need to even make general purpose robots. It’s not like individual humans practice every trade. You could have trade-specific robots. A drywall robot, a plumber robot, an electrician bot, etc. Each equipped only with those tools needed to complete just that trade. And instead of mechanical hands, they could simply be mounted.
I would agree that humanoid battery powered robots are not likely to result in mass employment in construction any time soon. But if you start relaxing those constraints, the possibility for mass worker displacement exists. A robotic construction site need not visibly resemble a human one.
The market. How many cable powered mobile bipedal robots have you seen?
Human workers drag power cables all over construction sites.
No they don’t? Have you ever been on a construction site? They have power cables, but they aren’t attached to them as they constantly move across a construction site. They are used for a specific job then rolled up and taken to the next job. Dozens of people dragging their own powered chords across a building site would not only be dangerous, but a huge pain in the ass.
but fixed industrial robots have been operating for many decades. The wear issues are obviously solvable in the right context.
As I said in my original statement, laypeople such as yourself tend to underestimate the amount of force a moving object creates. Just engaging in something like jogging can have a ground force reaction of 3x your body weight, something like a small jump can exert up to 7x your body weight. That is with more pliable triplaner joints covered in soft tissues that help reduce that ground reaction by absorbing some of the force with higher degrees of range of motion. Something more rigid like a robotic foot/ankle or prosthetics actually experience much higher degrees of ground reaction force.
They might not even have legs. A welding robot could just be a welding tool on the end of a long boom arm.
The entirety of my argument was around humanoid robots… That’s why I said it would make more sense for a robot on a flat factory floor to have wheels or tracks. I’m not arguing against the idea of robots in general, just bipedal humanoid ones.
Human workers drag power cables all over construction sites. Why can’t a robot be built that’s smart enough to manage its own power cord? Power cord management seems simple compared to doing meaningful construction work.
I’m sure it seems simple from your desk
One thing: do you think that those power cables are permanently connected to the workers? Are you thinking about workers using extension cords for tools or the ones laying cable for the building, because it sure as fuck isn’t simple to do either, which is why battery and petrol powered tools exist, and why the cableways are planned well in advance.
The wear and tear is a legitimate concern, but fixed industrial robots have been operating for many decades.
Only with maintenance and servicing by a human trained to see every potential flaw and monitor the equipment long term. Nothing, not even the most expensive equipment you can imagine, is without the need for regular maintenance.
You could have trade-specific robots. A drywall robot, a plumber robot, an electrician bot, etc. Each equipped only with those tools needed to complete just that trade. And instead of mechanical hands, they could simply be mounted.
Do you think that a construction site works the same way as a factory building cars? Just one long conveyors that shits out a fully built house? Workers move up, down and across the building dozens of times a day in a job doing multiple different jobs within their field, problem solving on the fly.
Interesting take. I guess I just realized every video we’ve ever seen of humanoid robots had been them doing something for a few minutes. But you never hear of one doing anything for days, months or years outside of science fiction.
Bipedal robots are kinda stuck in between a design flaw of what we want robots to look like, what we want robots to do, and what our technology is currently able to do.
Bipedal locomotion is just not the most efficient or simple mode of mobility for things like a flat factory floor, it’s really only efficient when you are factoring unleveled ground. We just currently don’t have the technology that allows bipedal efficiency across all terrain. That would require a synergistic system of optics and something that mimic propeoception. That’s a ton of information being sent from a vast network of sensory input being sorted out by your spine and brain in real time.
Imo robots for a factory floor should be on tracks or wheels. Attempting to design a robot that’s good for everything is just going to lead to a robot that is good for nothing. In engineering design should be made to fit a solution, designing for form and then trying to find a solution that form can solve is backwards and typically leads to terrible outcomes.
I’m in manufacturing and yeah it’ll be modified arm robots if anything. But current ones are large, heavy, and expensive. Cheaper than a person but they fuck up differently
Yep, in engineering design should be solution oriented. You don’t ever really want to first design the form and then try and see what solutions it may be able to solve.
A robot arm is a great start, you can then customize it further to your needs with different terminal end devices.
The main benefit of human hands is that they have an amazing amount of sensory input. If someone gets peripheral neuropathy and loses even a modest amount of that ability they can barely tie their shoes.
I remember reading a presser about how this cloud vendor was using robots for all kinds of tasks in their datacentre and it was super efficient because automation. But it never actually happened. They still have humans racking servers. The last 10% is hard.
They have a bunch of issues, and the only thing that is really automated about them is the walls. Since it’s basically unreinforced concrete they have a horrible time with insulation, cracking, delamination, and shrinkage. Not to mention that they aren’t very good at producing sharp or square angles, so a lot of your door and window work is custom or done manually.
If you know anything about residential construction, building the interior and exterior walls is not exactly an especially time consuming or expensive part of the build process. A decent sized team of good manual workers can put up the exterior and interior walls of a decent sized home in a week or two. The 3d house people say they can do it in 72 hours of machine time…which can run anywhere from 2 weeks to 5 months in real time.
you cpuld have a factory that cranks out walls and some sort of automated system that assembles them on site.
That idea is over a hundred years old. I currently live in a house that someone bought from a sears catalog in the 1930s. They would flat pack an entire house and ship it to you on a train for you to put together yourself, or you could hire a local sears approved contractor that had a team put it together for you.
IIRC “3D Printed” houses are falling apart due to lamination issues. That said, I think it’s a solvable problem, though it will take awhile to sort out the issues. Modular systems also show a lot of promise for mass production, potentially even of higher quality, and likely amenable to robotic assembly if carefully designed for it, but we know it’ll just be a profit seeking race to the bottom. Those construction jobs are safe for at least a decade, likely more.
IIRC “3D Printed” houses are falling apart due to lamination issues. That said, I think it’s a solvable problem, though it will take awhile to sort out the issues.
You’d probably have to invent a new type of concrete first. You can add resin binders to concrete to make wet concrete somewhat stick to drying concrete but it doesn’t work very well. The drying concrete tends to pull the moisture from the wet concrete which ends up interfering with the curing of both sides.
The main issue with 3d printing concrete is more of an inherent issue with how homes settle over time. Unless you are building on bedrock or you’re driving a massive amounts of piles for the foundation, structures will move and settle over time. This is why brick and mortar have been around for such a long time, the mortar gives the bricks enough flexibility to accommodate movement.
Concrete lacks this flexibility and will crack and crumble apart unless you reinforce it with rebar. Rebar gives concrete tensile strength, distributes weight across the whole structure, and holds cracked pieces together. So really any 3d printed home that isn’t reinforced somehow wont last very long.
In spite of the demonstration videos of robots running and jumping and carrying boxes and slowly sorting packages, none of them are anywhere close to being ready to handle construction jobs, or most manufacturing jobs. And, they can’t design or make the machines that automate the processes.
It’s a billionaire’s dream that this is happening now, it’s not. One of the major hurdles is power consumption, as we’re all seeing while they build out these “data centers”. Robots and computers cannot match the efficiency and self-sufficiency of a human. In a picture:
If people just stop eating food, we can use all the farmland for data centers and power plants
Yep. I work in prosthetics, people tend to overestimate current battery technology and underestimate how much energy it takes to move around a couple hundred pounds.
Even with batteries that cost a small fortune, we haven’t been able to create a lower limb powered prosthetic that can keep up with the human body for more than a couple hours.
There’s a lot of hype about the humanoid robotics coming out of China, but most everything is just marketing science fiction. Once you factor in the initial cost, power consumption for processing data, multiple batteries per unit, and the massive amount of general maintenance… It becomes pretty apparent that things aren’t going to pan out anytime soon.
I build limbs out of reinforced carbon fiber and titanium and even those acquire a ton of wear in a relatively small amount of time. Any bipedal robot with plastic parts, made to do intensive labour is going to start falling apart almost immediately. The human body takes a ton of damage throughout the day as well, but it’s self repairing.
The equation changes a lot when you start altering the assumptions.
Who says the robots have to have batteries? Human workers drag power cables all over construction sites. Why can’t a robot be built that’s smart enough to manage its own power cord? Power cord management seems simple compared to doing meaningful construction work.
The wear and tear is a legitimate concern, but fixed industrial robots have been operating for many decades. The wear issues are obviously solvable in the right context.
Keep in mind, prosthetics may not be the perfect analogy. We imagine these things as humanoid, but they need not be. You might have a drywall hanging robot that has suction cups to lift up panels and a screw gun to affix them. They might not even have legs. A welding robot could just be a welding tool on the end of a long boom arm.
There’s really no need to even make general purpose robots. It’s not like individual humans practice every trade. You could have trade-specific robots. A drywall robot, a plumber robot, an electrician bot, etc. Each equipped only with those tools needed to complete just that trade. And instead of mechanical hands, they could simply be mounted.
I would agree that humanoid battery powered robots are not likely to result in mass employment in construction any time soon. But if you start relaxing those constraints, the possibility for mass worker displacement exists. A robotic construction site need not visibly resemble a human one.
The market. How many cable powered mobile bipedal robots have you seen?
No they don’t? Have you ever been on a construction site? They have power cables, but they aren’t attached to them as they constantly move across a construction site. They are used for a specific job then rolled up and taken to the next job. Dozens of people dragging their own powered chords across a building site would not only be dangerous, but a huge pain in the ass.
As I said in my original statement, laypeople such as yourself tend to underestimate the amount of force a moving object creates. Just engaging in something like jogging can have a ground force reaction of 3x your body weight, something like a small jump can exert up to 7x your body weight. That is with more pliable triplaner joints covered in soft tissues that help reduce that ground reaction by absorbing some of the force with higher degrees of range of motion. Something more rigid like a robotic foot/ankle or prosthetics actually experience much higher degrees of ground reaction force.
The entirety of my argument was around humanoid robots… That’s why I said it would make more sense for a robot on a flat factory floor to have wheels or tracks. I’m not arguing against the idea of robots in general, just bipedal humanoid ones.
I’m sure it seems simple from your desk
One thing: do you think that those power cables are permanently connected to the workers? Are you thinking about workers using extension cords for tools or the ones laying cable for the building, because it sure as fuck isn’t simple to do either, which is why battery and petrol powered tools exist, and why the cableways are planned well in advance.
Only with maintenance and servicing by a human trained to see every potential flaw and monitor the equipment long term. Nothing, not even the most expensive equipment you can imagine, is without the need for regular maintenance.
Do you think that a construction site works the same way as a factory building cars? Just one long conveyors that shits out a fully built house? Workers move up, down and across the building dozens of times a day in a job doing multiple different jobs within their field, problem solving on the fly.
Interesting take. I guess I just realized every video we’ve ever seen of humanoid robots had been them doing something for a few minutes. But you never hear of one doing anything for days, months or years outside of science fiction.
Bipedal robots are kinda stuck in between a design flaw of what we want robots to look like, what we want robots to do, and what our technology is currently able to do.
Bipedal locomotion is just not the most efficient or simple mode of mobility for things like a flat factory floor, it’s really only efficient when you are factoring unleveled ground. We just currently don’t have the technology that allows bipedal efficiency across all terrain. That would require a synergistic system of optics and something that mimic propeoception. That’s a ton of information being sent from a vast network of sensory input being sorted out by your spine and brain in real time.
Imo robots for a factory floor should be on tracks or wheels. Attempting to design a robot that’s good for everything is just going to lead to a robot that is good for nothing. In engineering design should be made to fit a solution, designing for form and then trying to find a solution that form can solve is backwards and typically leads to terrible outcomes.
I’m in manufacturing and yeah it’ll be modified arm robots if anything. But current ones are large, heavy, and expensive. Cheaper than a person but they fuck up differently
Yep, in engineering design should be solution oriented. You don’t ever really want to first design the form and then try and see what solutions it may be able to solve.
A robot arm is a great start, you can then customize it further to your needs with different terminal end devices.
The main benefit of human hands is that they have an amazing amount of sensory input. If someone gets peripheral neuropathy and loses even a modest amount of that ability they can barely tie their shoes.
I remember reading a presser about how this cloud vendor was using robots for all kinds of tasks in their datacentre and it was super efficient because automation. But it never actually happened. They still have humans racking servers. The last 10% is hard.
Only because estimating the last 10% is so difficult.
In reality the last 10% is really at least half the job.
Don’t they have “3D Printed” houses?
I think the issue is, we are thinking of robots carrying boards and hammering nails etc.
Even not 3D printing you cpuld have a factory that cranks out walls and some sort of automated system that assembles them on site.
They have a bunch of issues, and the only thing that is really automated about them is the walls. Since it’s basically unreinforced concrete they have a horrible time with insulation, cracking, delamination, and shrinkage. Not to mention that they aren’t very good at producing sharp or square angles, so a lot of your door and window work is custom or done manually.
If you know anything about residential construction, building the interior and exterior walls is not exactly an especially time consuming or expensive part of the build process. A decent sized team of good manual workers can put up the exterior and interior walls of a decent sized home in a week or two. The 3d house people say they can do it in 72 hours of machine time…which can run anywhere from 2 weeks to 5 months in real time.
That idea is over a hundred years old. I currently live in a house that someone bought from a sears catalog in the 1930s. They would flat pack an entire house and ship it to you on a train for you to put together yourself, or you could hire a local sears approved contractor that had a team put it together for you.
IIRC “3D Printed” houses are falling apart due to lamination issues. That said, I think it’s a solvable problem, though it will take awhile to sort out the issues. Modular systems also show a lot of promise for mass production, potentially even of higher quality, and likely amenable to robotic assembly if carefully designed for it, but we know it’ll just be a profit seeking race to the bottom. Those construction jobs are safe for at least a decade, likely more.
You’d probably have to invent a new type of concrete first. You can add resin binders to concrete to make wet concrete somewhat stick to drying concrete but it doesn’t work very well. The drying concrete tends to pull the moisture from the wet concrete which ends up interfering with the curing of both sides.
The main issue with 3d printing concrete is more of an inherent issue with how homes settle over time. Unless you are building on bedrock or you’re driving a massive amounts of piles for the foundation, structures will move and settle over time. This is why brick and mortar have been around for such a long time, the mortar gives the bricks enough flexibility to accommodate movement.
Concrete lacks this flexibility and will crack and crumble apart unless you reinforce it with rebar. Rebar gives concrete tensile strength, distributes weight across the whole structure, and holds cracked pieces together. So really any 3d printed home that isn’t reinforced somehow wont last very long.