House-building robots and 3D-printing concrete buildings

October 15, 2018 12:00 PM

Japan's National Institute of Advanced Industrial Science and Technology (AIST) have developed a humanoid robot that can handle a variety of construction tasks when there's either a staffing shortage or serious hazards (Engadget). The prototype uses a mix of environment detection, object recognition and careful movement planning to install drywall by itself -- it can hoist up boards and fasten them with a screwdriver. If you prefer your homes made by a more mechanical-looking machine, here's how to 3D print an 800 square foot (~74.35 square meter) building in a day (Wired + short video from New Story & Icon), though roof, windows, doors and electrical/plumbing are installed with conventional methods.
posted by filthy light thief (37 comments total)

Ya but can it hide electrical boxes and cut off pigtails with a rotozip?
posted by Mitheral at 12:20 PM on October 15, 2018 [4 favorites]

I need a robot that buys the proper thickness of drywall to patch a hole in the ceiling, and can spackle, because apparently I can't do those things worth a damn myself.
posted by Foosnark at 12:21 PM on October 15, 2018 [1 favorite]

I was waiting for one of the Boston Dynamics robots to bust in and start smashing up the drywall.
posted by clawsoon at 12:44 PM on October 15, 2018 [9 favorites]

I'm curious - when these eventually deploy, will OSHA treat them workers and require it to wear PPE and high-vis, or are they a piece of equipment like a forktruck, and they'll all have a flashing light on top and a back-up chime?
posted by thecjm at 12:58 PM on October 15, 2018 [2 favorites]

Pretty interesting developments. I wonder if the robot would be better to just hold a heavy piece in place while humans fasten it.

The 3D printed house is cute, but the downside of electrical and plumbing having to go outside the walls is disappointing. But it's intended for people who don't have homes so for that aspect I expect it to work out pretty well.
posted by numaner at 1:07 PM on October 15, 2018

There is a lot of room for automation in construction. We are still using old methods for new assemblies and it makes very little sense- while certain older methods continue to need artisans to maintain quality (plaster, for example) we don’t need humans involved in erecting steel or running conduit.

Color me unimpressed on the “3D printed home”. That example image clearly has a separate steel structure holding up the roof. I guess “robot splorted walls” didn’t sound as nice.
posted by q*ben at 1:08 PM on October 15, 2018 [1 favorite]

from the Icon link which is a CBS news story, it sounds like it's designed to build the walls for a house that big in about 10 hours, then everything else still need to be put in, like the roof, utilities, doors & windows. But the goal is to build a row of these at a time in areas with a large homeless population. So ideally you can finish the walls and move onto the next plot while humans put in everything else.
posted by numaner at 1:25 PM on October 15, 2018

A company called Fastbrick Robotics floated by my attention a couple of years back-- their concrete block laying robot seems to be a great little system, not sure how far along it is to a commercial product, they had built a prototype at least.

Youtube video here of it here. One of the interesting things you can see is that because it knows what blocks go exactly where-- the blocks can be preloaded at the factory in that order-- if you pay attention to the rendering/simulation, you can see it leaving vertical conduit runs for cabling etc. Seems smart, and the technology is pretty much the same as surveying tools to maintain it's location.
posted by Static Vagabond at 1:37 PM on October 15, 2018 [3 favorites]

Drywall hanging, compare and contrast.
posted by Mei's lost sandal at 1:44 PM on October 15, 2018 [5 favorites]

But the goal is to build a row of these at a time in areas with a large homeless population

While I applaud their inventive spirit, the real problems are political, not technical, and so a technical solution will not be allowed to become a solution.

Back on topic, steel erection could be very highly automated (especially if you're doing a bolted frame; TCs are really good for this type of thing and I still think there's a way to get one-sided [no corresponding nut] connections using a TC-style bolt with specialist geometry but I can't quite work it out in my head yet. I'm discounting flow-drilling for a number of reasons I won't get into here).
posted by aramaic at 1:51 PM on October 15, 2018

Drywall hanging, compare and contrast.

Cool trick. Now compare and contrast an iPhone XS 256GB with a Tandy 1000 TX, a computer from one human generation ago -- and that cost twice as much, adjusted for inflation.
posted by lastobelus at 1:55 PM on October 15, 2018 [1 favorite]

These systems are prototypes, not even products yet. Judging their future based on what the prototypes can do is as inane as judging the future of personal computing based on what a Commodore Pet could do.
posted by lastobelus at 1:57 PM on October 15, 2018 [1 favorite]

im holding out for pure carbon houses laser printed out of graphene fog.
posted by mrjohnmuller at 2:40 PM on October 15, 2018 [2 favorites]

we don’t need humans involved in erecting steel or running conduit

I hate to be that cynic, but: humans aren’t going out of style until slave labor becomes more expensive than robots.

Not...not really a cause for celebration.
posted by schadenfrau at 2:51 PM on October 15, 2018 [1 favorite]

we don’t need humans involved in erecting steel or running conduit.

As someone who actually worked with a bunch of electricians last week to plan and implement a conduit run connecting four rooftop solar arrays on different roof planes to a common inverter down by the electric meter, I beg to differ.

The amount of on-the-fly problem-solving involved in construction is immense. Buildings are not these perfectly regular spherical-cow boxes that are amenable to careful planning and design. Nothing goes according to plan in construction. If you have the impression that it ever does, it's because the people you hired were good at their jobs and made it look easy. This is fieldwork, people.

Robots will have to get a lot smarter before they can handle construction. Self-driving cars are child's play compared to renovating a house. I'll trust AIs to write code first. Heck, I'll trust them to do surgery first.

Of course, it gets a lot easier if you take it out of the field and into a factory. That only works for new construction under a certain size though.

People who don't work in construction usually vastly underestimate how complex it is. I'm talking orders of magnitude, here.
posted by Anticipation Of A New Lover's Arrival, The at 3:06 PM on October 15, 2018 [10 favorites]

Ok, my dream tinyhouse design is now a spherical-cow box!

Autodesk started in the CAD/3D Max modeling space, like every cheesie 3D logo in the 90's. Currently their major business is taking the cad elements and generating the vast planning and bill of materials and architectural "stuff", basically online paperwork. Makes sense, going from an extensive cad design to accurate parts ordering is what computers do well. So walking by their research incubator one day, what is in the window: robot arms. Many different size and shapes of robot arms. So yep, not adding an extension to an old new england victorian this year, but some construction jobs will be automated in the next decade.
posted by sammyo at 3:20 PM on October 15, 2018 [1 favorite]

Also, that's not hanging drywall. That's hanging one sheet of drywall, under conditions that have been carefully engineered to be as easy as they can possibly be.

I really doubt that that thing is the future. The concrete-pooping snakebot? Maybe. But it's not going to come down to making generalist androids that can do exactly the same things as humans but better. It's going to come down to inventing totally new construction techniques that specialist, non-humanoid robots can excel at. Robots will never do what human construction workers currently do, because construction will change to meet robotics halfway.
posted by Anticipation Of A New Lover's Arrival, The at 3:22 PM on October 15, 2018 [3 favorites]

The important thing to understand about robots building houses is that robots don't need to sleep (caveat on charge, but... caveat). They don't need weekends. They don't need to see their kids soccer game. They don't have a baby on the way or get sick. They don't show up to work high. They don't knock one back too many at their lunch break. They don't take a lunch break. They just need as much uninterrupted space as necessary to minimize their collision recalculations, a steady supply of materials, and a sound proof enough environment that they can work 24/7 without causing problems with the neighbors.

We are at T₀ here folks. Going forward expect to see fewer construction jobs where the building is the same, pattern friendly and repeatable... That means these things are coming for the big steel office buildings first, but then they'll be elsewhere.

Is replacing a union worker with an capital hardware investment union busting?
posted by Nanukthedog at 3:30 PM on October 15, 2018

The other thing is, this is all new construction. There's a lot of pre-existing building stock that is going to need to be maintained pretty much indefinitely—and the older and weirder it is, the harder of a time robots are going to have working on it.
posted by Anticipation Of A New Lover's Arrival, The at 4:49 PM on October 15, 2018 [1 favorite]

There's a lot of pre-existing building stock that is going to need to be maintained pretty much indefinitely

*mumbles about rising sea levels, drowning coastal communities, climate change refugees needing housing, and increasingly severe weather that concrete would stand up to nicely*
posted by saysthis at 5:28 PM on October 15, 2018

There's a lot of pre-existing building stock

...interior refits would, I think, be pretty much bulletproof jobs for humans for quite a long time. You could possibly have a cost-effective machine handle the installation of light-gauge studs and track (I still doubt it, but I'll allow the possibility), but interior demo and installation of finishes will I suspect be fairly resistant to cost-effective mechanization.

As for electrical, not going to happen any time soon either -- except for overhead services ducting (as found inside hospital hallway ceilings); that's already being modularized and factory-built as rapidly as possible, and is also already fully laid-out in the MEP planning (um, assuming you're using competent contractors on your hospital job) so it's more amenable to mechanization. Maybe also hospital headwall assemblies too (although the gas-handling gives me pause).

...anyway, point being that I suspect most mechanization will be in those items that are in the process of being modularized and built off-site anyway. Those items have already shown that there is a cost benefit to handling them non-traditionally, and once you've got it being done in a factory, then you can more easily look at how to mechanize and automate the process without having to be excellent at everything on Day 1.
posted by aramaic at 5:53 PM on October 15, 2018

Anticipation, I work in construction. On my last job all of the steel and ductwork was prefabricated. All of the conduit was modeled in BIM down to 3/4”. There was plenty of field coordination, but I see no issues in moving to a greater use of automation, with field operator support, in many construction tasks. We are already seeing it. Total stations, point clouds and custom fabrication of light gauge and finish materials is all precursor knowledge.
posted by q*ben at 6:13 PM on October 15, 2018 [1 favorite]

Japan is in somewhat of a unique situation, experiencing a “super-aging” society both in rural and urban areas (Wikipedia). Japan needs more workers and it can't find them (CNN, 2017). Japan now has 1.48 jobs for every applicant. That's the highest number since 1974, when rapid growth pushed the ratio to 1.53.

"There's lots of people entering the labor market, which is one upside," said Marcel Thieliant, senior Japan economist at research firm Capital Economics.

The number of older people and women joining the workforce has increased because "the labor shortage is forcing companies to hire people who previously weren't looking for work," he said.

While the Japanese economy is growing, the labor market figures are more indicative of a shrinking pool of workers than a rise in the number of jobs, Thieliant said.

A rise in life expectancy and lower birth rates have created an aging population and dwindling workforce in Japan, posing a threat to the country's future economic growth. Japan is notoriously adverse to the idea of using immigration to offset the decline.

Emphasis mine, because while I can't speak to the veracity of this statement, it's a problematic one. But labor shortages aren't only in Japan, but also in central-eastern and north-western EU countries, where business surveys indicate that recent immigrants haven't taken local jobs.

If you're in the business (or academic field) of robotics, these labor shortages probably look like a problem to be solved, and citing "work in hazardous conditions" as an additional scenario to deploy robotic humanoids probably helps sell the idea to those skeptical about the universality of that labor shortage.

Too bad technology grants and venture capitalists aren't also interested in solving these "technical" problems by first looking at the regional to global social solutions that might actually solve the labor shortages more quickly, while also supporting people who currently have skills but lack the work. But I digress.
posted by filthy light thief at 7:41 PM on October 15, 2018 [2 favorites]

Construction has been experiencing the effects of machines and automation for over a century and there is no reason not to expect the curve to continue. Even brute manual labor today is a far cry from what it was decades ago. OSHA has done some good.

If my recollection serves, WPA projects were designed to be 90% labor and 10% materials. This was party to make-work, but partly just the realities of construction - especially heavy stone construction typified by those projects. Even consider a traditional Amish barn-raising. The labor input is pretty massive, even if compressed into a short time period by engaging the whole community.

Compare that to today's modern job site where whole elements such as a bathroom might come as a box, pre-plumbed in the factory. Right now that is primarily about labor relocation, to allow for better and faster manufacture in a reliable and controlled environment. But this is the sort of thing that could easily be automated. Today it modularity and prefabrication.

It isn't a far stretch to imagine that technology coming onto the job site. But I agree that for the most part it won't be humanoid robot type things. I suspect it will be work aids first. It's pretty easy to envision some sort of soft tracked vehicle that holds a vertical stack of dry wall, brings it from room to room for you, separates a sheet, lifts it and braces it against a wall for the human operator to verify location and apply fasteners. Is that a robot? Or a tool? I suspect we'll see the goalposts move on these definitions for a while.
posted by meinvt at 7:43 PM on October 15, 2018 [5 favorites]

On my walk in an older new england community I notice bits of architectural detail, bit of sculptured wood, swirls, small details on older houses the really give the environment a lovely feel. Contemporary building optimizes all that good stuff in the name of cost and efficiency. At some point 'smart' automation will allow those details and customization to return and all homes will be as beautiful or as quirky as one would like but could never afford due to the high cost of human labor.
posted by sammyo at 10:21 PM on October 15, 2018 [2 favorites]

I was going to post a video like the one of those rockers but then I didn't, I was going to complain about robot boosterism but than I thought better, but now, I skimmed the second video with the 3d printed masonry and fine, great, whatever but please do not try to sell a load of BS about it being a humanitarian endeavor to build housing for poor people. Poor people in the Global South, Third World, whatever you call it do not need a robot they need money for building materials and wages, they are completely able to build their own houses without a miracle from techworld.
posted by Pembquist at 10:21 PM on October 15, 2018 [6 favorites]

numaner: "The 3D printed house is cute, but the downside of electrical and plumbing having to go outside the walls is disappointing."

Lots of places where houses are constructed of brick/concrete and electrical is run in chases cut into the brick after the walls are built (this model cuts the complete groove not just two parallel cuts). It works as well as the N/A stick frame system.

Nanukthedog: " They just need as much uninterrupted space as necessary to minimize their collision recalculations, a steady supply of materials, and a sound proof enough environment that they can work 24/7 without causing problems with the neighbors."

Well, and maintenance. It's not like you can set these things loose and they'll keep going like energizer bunnies.
posted by Mitheral at 12:12 AM on October 16, 2018 [2 favorites]

I will concede that commercial new-build construction is probably the area most amenable to automation and modularization. But that's really what we're talking about here—automation. Automation doesn't need to mean fancy humanoid robots that do things just the same as human builders, in fact so far it's never meant that. It just means gradually making more and more capable machines, and gradually changing construction techniques to better utilize those machines' capabilities.

Even in residential construction, there are constantly new products coming out that aim to reduce the amount of manual labor involved in a task either through specialist tools or through materials and fittings that are designed to require fewer steps to install. For example, in my area the sheathing on houses has largely moved from CDX-plus-Tyvek to Advantech tongue-and-groove OSB with a water-resistant coating on one side. It's quicker to install because the tongue-and-groove system helps keep sheets aligned, and there's no wrapping step.

None of this involves androids, however.
posted by Anticipation Of A New Lover's Arrival, The at 2:21 AM on October 16, 2018 [1 favorite]

I thought the obvious solution for the concrete house would be to have two walls with a plenum in between. Run all your whatever inside the plenum (before the inner wall goes on) and then fill it all with sprayfoam or densepack. The inner wall needn't be concrete if that's not cost-effective; it could be regular old sheetrock with a wooden support structure, or maybe some kind of self-supporting foam insulation board with a finished face suitable for taping and mudding. In fact, probably better if it's *not* concrete because concrete is a huge pain to retrofit.
posted by Anticipation Of A New Lover's Arrival, The at 2:42 AM on October 16, 2018

I was about to add my own confident opinion, but then I remembered that people loudly and confidently proclaiming opposite predictions is a good sign that none of us know what we're talking about.

Maybe in twenty years we'll reach the robot singularity and they'll be designing themselves, or maybe we'll stagnate at robots not much more advanced than what we've got now. I do not know.

It is fun to make confident predictions, though. It's like being a sports fan declaring Exactly Who Will Win This Weekend And Why, but for technology.
posted by clawsoon at 6:12 AM on October 16, 2018 [1 favorite]

people loudly and confidently proclaiming opposite predictions is a good sign that none of us know what we're talking about.

Speak for yourself. Personally I haven't been wrong since 1982, which was two years before I was even born so I think I have a pretty good track record.
posted by Anticipation Of A New Lover's Arrival, The at 6:33 AM on October 16, 2018 [2 favorites]

as a person who spent years and years working on walking robots and humanoids, i can say i have no interest in having a robot build my house. and i also believe that reality is much further away than any of these articles suggest, having spent time both 1) making robots look good in a short video and 2) working to put a house together.

if i look at old buildings, i'm always amazed at craftsmanship and materials that age well. new buildings are always full of gross materials that look good for a year or two then age horribly.

rather than a robot built house, i'd love a return to human built houses that are smaller with a little more care taken... i'll just be over in the corner by myself with my romantic fantasies of a return to local communities and rewarding manual labor. ;)
posted by danjo at 7:04 AM on October 16, 2018 [5 favorites]

If you think of an excavator or a crane as a sort of exoskeleton, you could say that we've made great progress in that direction over the past century or so.
posted by clawsoon at 9:13 AM on October 16, 2018

At some point 'smart' automation will allow those details and customization to return and all homes will be as beautiful or as quirky as one would like but could never afford due to the high cost of human labor.

When everyone's super, no one will be.
posted by jon1270 at 4:24 AM on October 17, 2018

When everyone's super, no one will be.

By relative standards this is true, but by absolute standards it's false.
posted by clawsoon at 9:03 AM on October 17, 2018

hey hey no relitigation of The Incredibles!
posted by numaner at 12:23 PM on October 17, 2018

If you want plaster instead of drywall there's a plastering machine that appears to do a quick job of it.
posted by kgander at 7:29 PM on October 17, 2018 [1 favorite]