Evolving Floorplans • Hacker News Board

Re: Evolving Floorplans

#51

Post by ghein » Tue, Jul 31, 2018, 4:42 PM UTC

eadmund wrote: post #48

The author describes the results as 'irrational,' but I'd hardly agree with the negative connotation of that word: I think that the generated floorplans look lovely , and I wish that I'd gone to school somewhere like that. Far from irrational, I think it's quite rational to try to design buildings to be pleasant to live in, not just easy to draw with a straightedge. Also, the generated rooms look much like the sort o…

The reward function was horrible!

You need good sightlines for the kids, room for desks, and need to ensure kids aren't isolated in a corner. Plus light (not just to a small courtyard), ability to escape in case of fire or other emergency!!!!!!! and for minimal noise intrusion from other classrooms.

Minimal space and building material are just not the right constraints and speak to absolute ignorance of the actual problem on the part of the designer.

Re: Evolving Floorplans

#52

Post by hirundo » Tue, Jul 31, 2018, 4:43 PM UTC

The Hogwarts floorplan improves on this, reorganizing in real time to the needs of individual students and faculty. Any sufficiently advanced set of genetic algorithms, sensors and moving partitions is indistinguishable from magic.

Re: Evolving Floorplans

#53

Post by kuon » Tue, Jul 31, 2018, 4:45 PM UTC

All non square rooms I experienced where terrible to place furniture.

Re: Evolving Floorplans

#54

Post by foolfoolz » Tue, Jul 31, 2018, 4:45 PM UTC

kickopotomus wrote: post #42

foolfoolz wrote: post #40

i am a product of american public school. in every school all the classrooms (except portables) had a wall that was mostly windows. you can see thats also true in the original school design. i’d rather have a ton of windows than shorter distance to walk. you could feel the enclosed difference in portable rooms

He talks about windows if you scroll down.

yea i saw the whole thing. the courtyards won’t feel the same. you’ll just have one small wall opening up to it

Re: Evolving Floorplans

#55

Post by hammock » Tue, Jul 31, 2018, 4:48 PM UTC

These types of exercises are useful not to find their solutions, but to reveal the shortcomings of the model used to create them - and our thus our own mental models!

Building design is older than humankind, and there's tens of millenia worth of reasons why they are the way they are. But those reasons aren't always clear. Essentially modern floor plan design is the output of a black-box machine (human) learning algorithm.

The flawed output here, and the comments, help us point out what some of the missing variables are: egress, airflow, construction cost, etc. Do enough of this and you can get a vastly improved model.

Re: Evolving Floorplans

#56

Post by theoh » Tue, Jul 31, 2018, 4:49 PM UTC

jethro_tell wrote: post #50

TeMPOraL wrote: post #37

Earlier quoted context omitted.

What about concrete 3D-printing? (I'm not even sure if it's real or feasible; I just remember a bunch of pop news articles with a video of a large CNC machine depositing concrete to construct a building.)

Concrete is manual 3D printing from 100+ years ago. =) I've seen a concrete 3d printer, but it doesn't look like it would be structural. i.e. not sure how you would print around full lengths of rebar?

You have to build formwork to construct with in-situ concrete, so there's still carpentry (or equivalent) involved.

It's a good point that existing architectural concrete 3d printing creates "mass concrete" not reinforced concrete, but mass concrete can be structural.

Re: Evolving Floorplans

#57

Post by pc86 » Tue, Jul 31, 2018, 4:49 PM UTC

lainga wrote: post #45

Right optimized one looks best, because it's the only one that doesn't squirrel any of the big rooms (library, gym, cafeteria) away at the ends of corridors.

I think it's probably a balancing act between placing the big room at the end of a hallway (making that hallway a nightmare prior to any event) and placing it in the middle of the building (making that intersection a nightmare prior to any event). I'm sure it depends largely on the specifics of the room and how often it will be used of course. Less frequently utilized is probably better off further away from the center, more frequently better off closer but with multiple entrances & exits to ease congestion.

Re: Evolving Floorplans

#58

Post by pavel_lishin » Tue, Jul 31, 2018, 4:50 PM UTC

ggg9990 wrote: post #32

pavel_lishin wrote: post #15

Earlier quoted context omitted.

There are actually a few good reasons beyond "ze children must be OPTIMIZED!" - fire safety, mentioned in the post, is one of them.

How many elementary school children have perished in school fires in the last 25 years?

Thankfully, zero: https://www.nfpa.org/News-and-Research/Fire-statistics-and-r...

But your point is well taken; there's no particularly great reason to, unless your school is so big and your administration is so goofy that kids are constantly running late. That constraint would be better suited to applying these algorithms to offices, hospitals, military bases, prisons, and maybe shopping malls.

Re: Evolving Floorplans

#59

Post by theoh » Tue, Jul 31, 2018, 4:54 PM UTC

tw1010 wrote: post #5

Sure, it looks cool. But there are so many other properties of a home that aren't accounted for in this model (mostly aesthetic ones, like cozyness, feelings of safety/comfort, how the plan interacts with light, etc). One day we'll live in homes with automated floorplans, I'm sure of it. But it won't happen (or at least, we won't be happy about living in such things) until the day we figure out ways to deal with more…

Why are you sure that at some point in the future, we will live in homes with AI-generated floorplans?

It's a bit like saying that poetry will mostly be written by computers at some point in the future, or humans will mostly have sex with robots in the future. There's going to be more to this than whether the aesthetic experience is acceptable.

Re: Evolving Floorplans

#60

Post by vortico » Tue, Jul 31, 2018, 4:57 PM UTC

ggg9990 wrote: post #32

pavel_lishin wrote: post #15

Earlier quoted context omitted.

There are actually a few good reasons beyond "ze children must be OPTIMIZED!" - fire safety, mentioned in the post, is one of them.

How many elementary school children have perished in school fires in the last 25 years?

https://www.nfpa.org/public-education/by-topic/property-type...

>U.S. fire departments responded to an estimated average of 4,980 structure fires in educational properties in 2011–2015, annually. These fires caused annual averages of one civilian death, 70 civilian injuries, and $70 million in direct property damage. Nearly seven in ten of the fires in educational properties (69%) occurred in nursery, elementary, middle, or high

So roughly 25 deaths and 1750 injuries in the last 25 years in the US. I think this is amazingly low considering the 4,980 structure fires in schools per year.