Hod Lipson: Robots that are "self-aware"

Prevoditelj: Tilen Pigac - EFZG Recezent: Mislav Ante Omazić - EFZG

So, where are the robots?

Dakle, što su roboti?

We've been told for 40 years already that they're coming soon.

Već nam se 40 godina priča kako će se uskoro pojaviti.

Very soon they'll be doing everything for us.

Ubrzo oni će raditi sve za nas:

They'll be cooking, cleaning, buying things, shopping, building. But they aren't here.

oni će kuhati, čistiti, kupovati stvari, ići u kupovinu, graditi. Ali oni nisu ovdje.

Meanwhile, we have illegal immigrants doing all the work,

U međuvremenu, imamo ilegalne imigrante koji rade sav taj posao,

but we don't have any robots.

ali nemamo robote.

So what can we do about that? What can we say?

Dakle, što možemo učiniti oko toga? Što možemo reći?

So I want to give a little bit of a different perspective

Dakle, želim dati malo drugačiju perspektivu

of how we can perhaps look at these things in a little bit of a different way.

o tome kako možda možemo gledati na te stvari na pomalo drugačiji način.

And this is an x-ray picture

A ovo je rendgenska snimka

of a real beetle, and a Swiss watch, back from '88. You look at that --

prave bube i švicarskog sata, iz 1988.godine. Gledate u to --

what was true then is certainly true today.

ono što je bilo istinito tada je definitivno istinito danas.

We can still make the pieces. We can make the right pieces.

Mi još uvijek možemo izraditi dijelove, mi možemo izraditi prave dijelove,

We can make the circuitry of the right computational power,

mi možemo izraditi strujne krugove prave računalne moći,

but we can't actually put them together to make something

ali mi ih zapravo ne možemo staviti zajedno kako bi napravili nešto

that will actually work and be as adaptive as these systems.

što će zapravo raditi i biti prilagodljivo poput ovih sustava.

So let's try to look at it from a different perspective.

Dakle, pokušajmo ih gledati iz druge perspektive.

Let's summon the best designer, the mother of all designers.

Pozovimo najboljeg dizajnera, majku svih dizajnera:

Let's see what evolution can do for us.

da vidimo što evolucija može učiniti za nas.

So we threw in -- we created a primordial soup

Dakle, ubacili smo -- stvorili smo praiskonsku juhu

with lots of pieces of robots -- with bars, with motors, with neurons.

s mnogim dijelovima robota: sa šipkama, s motorima, s neuronima.

Put them all together, and put all this under kind of natural selection,

Stavite ih sve zajedno, i stavite sve to pod jednu vrstu prirodne selekcije,

under mutation, and rewarded things for how well they can move forward.

pod mutaciju, i nagrađene stvari ovisno o tome kako dobro se mogu kretati naprijed.

A very simple task, and it's interesting to see what kind of things came out of that.

Jako jednostavan zadatak, i zanimljivo je vidjeti kakve vrste stvari su proizašle iz toga.

So if you look, you can see a lot of different machines

Ako pogledate, možete vidjeti kako je mnogo različitih strojeva

come out of this. They all move around.

proizašlo iz toga. Svi se oni kreću okolo,

They all crawl in different ways, and you can see on the right,

svi oni pužu u različitim smjerovima, i možete vidjeti ovdje na desnoj strani,

that we actually made a couple of these things,

da smo zapravo napravili par tih stvari,

and they work in reality. These are not very fantastic robots,

i one rade u stvarnosti. Ovo nisu jako fantastični roboti,

but they evolved to do exactly what we reward them for:

ali su evoluirali do toga da rade točno ono za što ih nagrađujemo:

for moving forward. So that was all done in simulation,

za kretanje unaprijed. Dakle, to je sve napravljeno pomoću simulacije,

but we can also do that on a real machine.

ali možemo to učiniti i na pravom stroju.

Here's a physical robot that we actually

Ovdje je fizički robot kojem smo zapravo

have a population of brains,

usadili mozak,

competing, or evolving on the machine.

koji se takmiči, ili evoluira, na stroju.

It's like a rodeo show. They all get a ride on the machine,

To je poput rodea: svi oni dobiju vožnju na stroju,

and they get rewarded for how fast or how far

i svi oni budu nagrađeni ovisno kako brzo ili kako daleko

they can make the machine move forward.

mogu natjerati stroj da se kreće naprijed.

And you can see these robots are not ready

I možete vidjeti da ti roboti nisu spremni

to take over the world yet, but

da preuzmu svijet još, ali

they gradually learn how to move forward,

oni postepeno uče kako se kretati naprijed,

and they do this autonomously.

i to rade samostalno.

So in these two examples, we had basically

Dakle, u ova dva primjera, mi smo u osnovi imali

machines that learned how to walk in simulation,

strojeve koji su učili kako hodati u simulaciji,

and also machines that learned how to walk in reality.

i isto tako strojeve koji su učili kako hodati u stvarnosti.

But I want to show you a different approach,

Ali želim vam pokazati drugačiji pristup,

and this is this robot over here, which has four legs.

a to je ovaj robot, ovdje, koji ima četiri noge,

It has eight motors, four on the knees and four on the hip.

ima osam motora, četiri na koljenima i četiri na boku.

It has also two tilt sensors that tell the machine

Ujedno ima i dva nagibna senzora koji govore stroju

which way it's tilting.

u kojem smjeru je nagib.

But this machine doesn't know what it looks like.

Ali ovaj stroj ne zna kako izgleda.

You look at it and you see it has four legs,

Gledate u njega i vidite da ima četiri noge,

the machine doesn't know if it's a snake, if it's a tree,

stroj ne zna je li to zmija, ili drvo,

it doesn't have any idea what it looks like,

nema nikakvu ideju o tome kako izgleda,

but it's going to try to find that out.

ali pokušat će to doznati.

Initially, it does some random motion,

Inicijalno, radi neke nasumične pokrete,

and then it tries to figure out what it might look like.

i zatim pokušava dokučiti na što bi mogao ličiti --

And you're seeing a lot of things passing through its minds,

i vidite mnogo stvari koje prolaze kroz njegove misli,

a lot of self-models that try to explain the relationship

mnogo samo-modela koji pokušavaju objasniti vezu

between actuation and sensing. It then tries to do

između stavljanja u pokret i osjećanja -- i zatim pokušava uraditi

a second action that creates the most disagreement

drugu radnju koja stvara najviše neslaganja

among predictions of these alternative models,

između predviđanja tih opcijskih modela,

like a scientist in a lab. Then it does that

poput znanstvenika u laboratoriju. Zatim radi to

and tries to explain that, and prune out its self-models.

i pokušava to objasniti, i izrezati vlastite samo-modele.

This is the last cycle, and you can see it's pretty much

Ovo je posljednji ciklus, i možete vidjeti da je više-manje

figured out what its self looks like. And once it has a self-model,

dokučio kako njegovo biće izgleda, i jednom kada ima samo-model,

it can use that to derive a pattern of locomotion.

može to iskoristiti da izvuće uzorak kretanja.

So what you're seeing here are a couple of machines --

Dakle, ono što vidite ovdje je par strojeva --

a pattern of locomotion.

uzorak kretanja.

We were hoping that it wass going to have a kind of evil, spidery walk,

Nadali smo se kako će imati tu neku vrstu zlobnog, paukovskog hoda,

but instead it created this pretty lame way of moving forward.

ali umjesto toga, stvorio je prilično jadan način kretanja naprijed.

But when you look at that, you have to remember

Ali kada to gledate, morate upamtiti

that this machine did not do any physical trials on how to move forward,

kako taj stroj nije radio nikakve fizičke pokuse kako se kretati unaprijed,

nor did it have a model of itself.

niti je imao model samog sebe.

It kind of figured out what it looks like, and how to move forward,

Nekako je sam dokučio kako izgleda, i kako se treba kretati naprijed,

and then actually tried that out.

i zatim je zapravo to pokušao.

(Applause)

(Pljesak)

So, we'll move forward to a different idea.

Dakle, mi ćemo krenuti na drugačiju ideju.

So that was what happened when we had a couple of --

Dakle, to se dogodilo kada smo imali par --

that's what happened when you had a couple of -- OK, OK, OK --

to se dogodilo kada si imao par -- OK, OK, OK --

(Laughter)

(Smijeh)

-- they don't like each other. So

-- ne vole jedan drugoga. Dakle,

there's a different robot.

ovdje je različit robot.

That's what happened when the robots actually

To je ono što se dogodilo kada roboti zapravo

are rewarded for doing something.

budu nagrađeni za nešto što su napravili.

What happens if you don't reward them for anything, you just throw them in?

Što se događa ako ih ne nagradite za bilošto, već ih samo ubacite unutra?

So we have these cubes, like the diagram showed here.

Dakle, imamo te kocke, kako je to ovdje prikazano na dijagramu.

The cube can swivel, or flip on its side,

Kocka se može okretati, ili se preokrenuti na stranu,

and we just throw 1,000 of these cubes into a soup --

i mi samo ubacimo 1.000 takvih kocki u juhu --

this is in simulation --and don't reward them for anything,

ovo je u simulaciji -- i ne nagradimo ih za išta,

we just let them flip. We pump energy into this

jednostavno im damo da polude. Upumpavamo energiju u to

and see what happens in a couple of mutations.

i vidimo što se događa u par mutacija.

So, initially nothing happens, they're just flipping around there.

Dakle, inicijalno, ništa se ne događa, samo lude tamo.

But after a very short while, you can see these blue things

Ali nedugo nakon toga, možete vidjeti ove plave stvari

on the right there begin to take over.

na desnoj strani koje počinju preuzimati.

They begin to self-replicate. So in absence of any reward,

Počinju se samo-razmnožavati. Dakle, u odsustvu ikakve nagrade,

the intrinsic reward is self-replication.

intrinzična nagrada je samo-razmnožavanje.

And we've actually built a couple of these,

I mi smo zapravo izradili par tih,

and this is part of a larger robot made out of these cubes.

i to je dio većeg robota koji je napravljen od tih kocki,

It's an accelerated view, where you can see the robot actually

to je ubrzan prikaz, gdje možete vidjeti kako robot zapravo

carrying out some of its replication process.

prolazi kroz neki od procesa razmnožavanja.

So you're feeding it with more material -- cubes in this case --

Dakle, hranite ga s više materijala -- kocki u ovom slučaju --

and more energy, and it can make another robot.

i više energije, i može stvoriti još jedan robot.

So of course, this is a very crude machine,

Dakle, naravno, ovo je vrlo sirov stroj,

but we're working on a micro-scale version of these,

ali radimo na mikro-verziji njih,

and hopefully the cubes will be like a powder that you pour in.

i nadamo se kako će te kocke biti poput praha koji ulijete.

OK, so what can we learn? These robots are of course

U redu, dakle, što možemo naučiti? Ti roboti naravno

not very useful in themselves, but they might teach us something

nisu sami po sebi korisni, ali bi nas mogli naučiti nešto

about how we can build better robots,

o tome kako možemo izraditi bolje robote,

and perhaps how humans, animals, create self-models and learn.

i možda kako ljudi, životinje, stvaraju samo-modele i uče.

And one of the things that I think is important

I jedna od stvari za koju smatram da je važna

is that we have to get away from this idea

da se moramo odmaknuti od te ideje

of designing the machines manually,

ručnog dizajniranja strojeva,

but actually let them evolve and learn, like children,

već im dopustiti da evoluiraju i uče, poput djece,

and perhaps that's the way we'll get there. Thank you.

i možda je to način da stignemo tamo. Hvala vam.

(Applause)

(Pljesak)