Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Where are the robotic bricklayers?

constructionphysics.substack.com

Post reply on HN
Print view View original

461–470 of 564 posts

Re: Where are the robotic bricklayers?

#461

Post by eru »

h0l0cube wrote: post #282
eru wrote: post #274

Earlier quoted context omitted.

I'm not sure classic procedural content generation like in rogue-likes is all that comparable to using a GAN to run the whole game? Have a look at the video, it's quite impressive.

I've seen it before, and I've a basic understanding of GANs, I just don't see it being overly useful. This technique can make a really blurry simulacrum of an actual game, and that's really cool, but I'm not sure how it could be used to make a something both truly novel and coherent. There's plenty of low hanging fruit for AI within an engine, whereas using AI to be the entire engine is somewhat infeasible

You are right about the technique not being very useful as of today. My fascination stems from my assumption that more resources poured into this approach would yield vastly better results.

Even just watching the video, I came up with several possible improvements to try out. Eg adversarial training, that would really hone in on the situations and aspects where the model is weak so far, like edge conditions; instead of just using normal gameplay as input.

Re: Where are the robotic bricklayers?

#462

Post by eru »

esailija wrote: post #446
eru wrote: post #276

Earlier quoted context omitted.

So far Google's self-driving cars have vastly fewer accidents than humans have per mile. Or what kind of performance are you interested in? I am sure, they could also be made to drive faster than humans and still be safer on average thanks to superior reflexes and foresight. But what do you mean by 'across the board'? As far sources, a web search gives many articles about safety of self driving cars. See eg https://w…

You are probably comparing accidents of self-driving cars in optimal conditions (since they cannot even drive in any other conditions lol) to all accidents in all conditions in human drivers.

Humans get into plenty of accidents under clear skies, too.

See https://www.forbes.com/sites/bradtempleton/2020/10/30/waymo-... which tries somewhat to correct for conditions.

Re: Where are the robotic bricklayers?

#463

Post by Chris2048 »

spyckie2 wrote: post #31

The value of work is not in the work done; it's in the decisions being made. It's the ability to handle edge cases, to work your way through ambiguity, to "unstuck" a stuck situation, and to resolve tricky unforeseen issues. If you build a robot that only handles the main workflow and can't handle any of the exceptions, you're not automating anything; you're building a convenience tool. Basically, you're removing som…

> economically unviable

> You're not reducing labor costs by much

You could make the same argument wrt supermarket self checkout. They still need a worker to manage the machines. Yet you need fewer human workers overall, so productivity still increases per-worker at the right scale. Allow the tech to develop and they may become more viable.

arguing "you still need humans" seems to me to be a straw-man; there no rule that this is useless so long as humans are still involved - if productivity increases, that's enough.

> you're building a convenience tool

Which seems fine to me. Removing the majority of the "labour" part of the job removes the most unskilled part that humans are least appropriate for (repetitive uncreative/boring work) and freeing them up for the "hard/creative decisions" aspect that humans are better at.

> largely unpopular

> workers lose control over their job

This might be unpopular among workers; not sure, other worker might prefer it. At least, if this requires fewer workers to lay brick, either the demand is such that it can still meet worker supply, or workers suffer (from lower wages or unemployment).

But this is irrelevant - barring union influence, it's not the workers that get to decide. If the job is to "babysit machines" you either refuse the job, or you don't - so long as people take the job, they will grow.

Re: Where are the robotic bricklayers?

#464

Post by himinlomax »

necovek wrote: post #404
MillenialMan wrote: post #394

Earlier quoted context omitted.

I don't really understand this. When I see a small house being built, it will have four or five builders on-site at all times, mostly doing labour (and they're, frankly, extremely slow. It will take months to lay the bricks). If you still need the decisions, can't you cut that down to one, or maybe two? That strikes me as a dramatic reduction in labour cost.

Multi-story buildings take only a few days of brick-laying by 4-5 people. Before you can lay the bricks though, you need a firm foundation to carry those bricks, including those concrete floors which take a while to dry up. There is still a bunch of work in construction that can happen while brick laying happens (plumbing and electrical installations in the floor, pillars to carry roof/next floor, ...) and there is u…

> including those concrete floors which take a while to dry up.

FWIW, the slower concrete dries, the stronger it gets. To the point where critical concrete buildings like bridges are kept moist for days or even weeks with hoses and wet blankets.

Re: Where are the robotic bricklayers?

#465

Post by himinlomax »

function_seven wrote: post #146
dragonsky67 wrote: post #130

The question I would have around automated building construction is "why use bricks?". Bricks are great for humans to build with, they are just the right size and weight for the human hand to manipulate and place. They are not optimised for machine manipulation. I would guess that if you wanted to automate construction you would start by making things like walls on an automated production line, then ship the largest…

Because they’re pleasant to look at. Half the time, the wall isn’t even “needed”, other than for aesthetic purposes. Whenever a wall is purely functional, it’s almost never made of small clay bricks today. I’m talking about foundation walls, subsurface retaining walls, and similar things. The types of walls that aren’t meant to be looked at, but instead meant to hold something back or up. Whenever you see a wall bein…

> Because they’re pleasant to look at.

The article mentions cinder blocks ("CMU") -- those are not typically pleasant to look at. Even the ones that are less unpleasant are usually covered, with sidings or paint.

Re: Where are the robotic bricklayers?

#466

Post by esailija »

sharkweek wrote: post #208
NoPicklez wrote: post #203

Earlier quoted context omitted.

And sitting in an office chair in front of a computer all day doesn't bring its own health problems? Weight gain due to limited daily movement, forward posture due to sitting and looking forward all day, RSI due to typing and mouse movement.

A question for the crowd, then: what is the “best” job for the human body? Because sitting at a desk all day for me definitely isn’t it.

Well unfortunately it doesn't get better than that, except when working as remote you can lie on a sofa or bed. After a full day you still have full energy to do optimal amount of physical exercise, just enough (which is relatively little) to stay in optimal condition without destroying your body.

Re: Where are the robotic bricklayers?

#467

Post by moduspol »

SlapperKoala wrote: post #422
munificent wrote: post #93

With human bricklayers, you can find enough people who have that skill that the pay is low. Here in the US, the availability of illegal immigrants means that the pay is often literally criminally low and you don't have to provide benefits. Replace those bricklayers with a robot and its caretakers and now you need: 1. Someone with technical training who can keep the bot running happily. Since the technology is new and…

The cheapest way to automate something is always poor people.This applies even more so in the developing world. Where the vast majority of new construction is happening. Though it would be interesting to see what the state of construction is like in developed countries that don't have access to a vast pool of cheap immigrant labor. Maybe New Zealand? Northern Europe?

Not everywhere in the US has a vast pool of cheap immigrant labor. That's a fairly localized thing.

I'm just outside of Pittsburgh and, I guess I can't say conclusively, but the vast majority of landscapers, construction workers, and other outdoor laborers appear to come from the general local population. It's not like what I hear about California / Texas where hispanic immigrants are disproportionately represented.

Re: Where are the robotic bricklayers?

#468

Post by iandanforth »

Important insights from this article:

- The hard part is the mortar

- Any useful number of brick lays will require after placement adjustment to maintain level/true alignment

- Those adjustments are where a good deal of 'skill' comes in

- Finishing / jointing is even harder

So it looks like this is an RL problem and the focus of the research needs to be geared toward the above. Basically you need a system to understand mortar and how to place various shapes of objects in it so that over time you achieve the goal of a specific wall shape.

This problem suggests an arm-farm like setup. Normally I always urge sim-first, but I'm not aware of any useful simulator that can do non-Newtonian fluids and robotic systems well. (either sure, but both? Let me know if you know of one.)

Re: Where are the robotic bricklayers?

#469

Post by papito »

I mean, there are already prototypes: https://www.youtube.com/watch?v=MVWayhNpHr0

It's only a matter of time. A building is planned, the step-by-step orchestration is QA'ed by a qualified engineer, and a set of robots go to work to build a whole building. Construction jobs will be phased out.

Also, in the movie Runaway - which actually predicted police areal drones as well, almost 30 years ago (but the concept of a drone did not exist yet, so it was called a "floater").

Re: Where are the robotic bricklayers?

#470

Post by Chris2048 »

MomoXenosaga wrote: post #389
atoav wrote: post #300

Earlier quoted context omitted.

Yeah, but houses are rarely like a serial manufactured car. Sometimes you have to lay down every row of bricks differently with a ton of unplanned (or unplannable) things to account for, because the electrician might work while you lay your brick etc. Not that you cannot automate whole houses and sell them off the shelve for cheap (see prefabs), it just isn't something that could be done at a site where your brickbot…

In my country houses are build in factories and then assembled on site. You rarely build a single house you build an entire neighbourhood in one go.

Are the factories mobile and/or relocatable to reduce transport costs; and/or are the neighbourhood location requirements somewhat specific (e.g. flat stable land, good road access etc)?
Post reply on HN