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

Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

nature.com

Post reply on HN
Print view View original

91–100 of 117 posts

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#91

Post by TeMPOraL »

nl wrote: post #70
seandavidfisher wrote: post #67

> Previous studies have found that dandelion seeds always have between 90 and 110 bristles, says Nakayama Those poor research assistants. Imagine counting hundreds of dandelion bristles every day. Probably still not possible with AI/image recognition either.

That would 100% be possible with image recognition. Segment using UNet, then measure the remaining joined pixels. (Well thanks for the downvotes. I do this as my job, so I guess I'm doing the impossible or something)

Hm. Are NNs reliable enough to be sure you're getting a correct result?

I'd imagine this kind of task would lend itself to some old-school image recognition techniques - photograph against uniform background, threshold, mask the middle part and count contiguous regions.

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#92

Post by kabacha »

comex wrote: post #80
mdda wrote: post #76

Earlier quoted context omitted.

Perhaps people are wondering how your UNet is going to see the bristles on the other side of the dandelion seed.

Take photos from multiple angles?

And how do you deduplicate the ones that are visible on multiple angles?

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#93

Post by hasahmed »

im3w1l wrote: post #64

If you find this interesting I highly recommend you to look at some vape-trick videos. Smoke/steam is great at visualizing vortices, and it's absolutely astonishing how much aerodynamically control some people can achieve. Creating lots of small vortices in rapid succession, merging vortices, splitting vortices, letting vortices "suck in" other vortices. https://www.youtube.com/watch?v=Tmv228G8R4o&time_continue=1m

I feel like I could have enjoyed that if those dudes weren't such giant fuck boys

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#94

Post by coldtea »

mindfulhack wrote: post #31

It's like nature's had millions of years to crunch through its own AI / machine learning of evolution and adaptation, and we're about to unleash the same thing on an impossibly faster scale. Imagine the technological breakthroughs that are going to happen. I'm optimistic.

>Imagine the technological breakthroughs that are going to happen.

Or the horrific, ending-of-human-race potential for mistakes (or purposeful harmful action for profit/terrorism/state actions/etc).

I'm realistic.

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#95

Post by coldtea »

paavoova wrote: post #48

Perhaps ignorance on my part, but: why do they require highspeed cameras and laser illumination to figure this out in 2018? Shouldn't physics by able to model such a relatively simply structure and how air would move through and around it?

"Relatively simply structures" can be nigh impossible to model...

https://en.wikipedia.org/wiki/Three-body_problem

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#96

Post by TeMPOraL »

aidos wrote: post #3

That was a lovely little diversion! There’s a brief 1 minute clip in the article that’s worth watching. I do have one question though, how is the air flowing up through? It the seed not falling for that to happen?

I imagine they get blown with the wind, start falling, which triggers the formation of the vortex - and that vortex is somewhat self-sustaining, almost completely arresting the fall.

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#97

Post by titzer »

nl wrote: post #70
seandavidfisher wrote: post #67

> Previous studies have found that dandelion seeds always have between 90 and 110 bristles, says Nakayama Those poor research assistants. Imagine counting hundreds of dandelion bristles every day. Probably still not possible with AI/image recognition either.

That would 100% be possible with image recognition. Segment using UNet, then measure the remaining joined pixels. (Well thanks for the downvotes. I do this as my job, so I guess I'm doing the impossible or something)

ffs not everything needs ML. It's not like they need to scale to every individual dandelion in the known universe.

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#99

Post by pbhjpbhj »

apo wrote: post #6

Amazing that: (1) the mechanism involves a detached vortex; and (b) that it's taken millennia to understand what makes dandelion seeds fly.

Re (b) they appear to "fly" just like any other small piece of stuff, just moving with the air, so I imagine it's more "there doesn't appear to be any need to explain it".

They, dandelion seeds, appear to fly like other cotton-y plants do (eg heather [1]). I wonder if they creates these vortices too?

I'd guess not as their structures aren't regular and they don't have the weighted 'drop' to give stability?

[1] https://oregonstate.edu/dept/nursery-weeds/email_pubs/thistl...

Re: Dandelion Seeds Fly Using ‘Impossible’ Method Never Before Seen in Nature

#100

Post by pbhjpbhj »

namirez wrote: post #28
mattdemon wrote: post #17

Wow, new parachute designs may be possible then! They will look scary, but can be more effective.

Probably not! This phenomenon is not scale-independent. There is a parameter called the Reynolds number (Re for short) which is the ratio of intertial forces to viscous forces. For a dandelion seed Re is small which is the key to the stability of the vortex. For a parachute, the Re number is much higher which makes the dynamics of the flow chaotic (called turbulence). There is a critical Re number beyond which There…

Perhaps many small dandelion apertures could be used in a larger chute (not necessarily the whole surface), rather than trying to make a single large seed-chute.

Perhaps I should have tried to patent that ...

Post reply on HN