David Hanson: Robots that "show emotion"

187,811 views ・ 2009-10-14

TED


Please double-click on the English subtitles below to play the video.

Prevodilac: Jelena Nedjic Lektor: Ana Zivanovic-Nenadovic
00:16
I'm Dr. David Hanson, and I build robots with character.
0
16260
3000
Ja sam dr Dejvid Henson i ja pravim robote sa karakterom.
00:19
And by that, I mean
1
19260
2000
Pri tome mislim
00:21
that I develop robots that are characters,
2
21260
2000
da stvaram robote određenog karaktera,
00:23
but also robots that will eventually
3
23260
3000
ali isto tako i robote koji će na kraju
00:26
come to empathize with you.
4
26260
2000
saosećati sa vama.
00:28
So we're starting with a variety of technologies
5
28260
2000
Počeli smo sa primenom različitih tehnologija
00:30
that have converged into these conversational character robots
6
30260
4000
koje su zajedno primenjene kod ovih robota sa karakterom koji govore,
00:34
that can see faces, make eye contact with you,
7
34260
2000
koji mogu videti lica, uspostavljaju konakt očima sa vama,
00:36
make a full range of facial expressions, understand speech
8
36260
3000
ispoljavaju puni spektar facijalnih ekspresija, razumeju govor
00:39
and begin to model how you're feeling
9
39260
4000
i počinju da modeluju kako se osećate
00:43
and who you are, and build a relationship with you.
10
43260
3000
i ko ste vi, i mogu da izgrade odnos sa vama.
00:46
I developed a series of technologies
11
46260
2000
Razvio sam čitav serijal tehnologija
00:48
that allowed the robots to make more realistic facial expressions
12
48260
3000
koje omogućuju ovim robotima da prave uverljivije izraze lica
00:51
than previously achieved, on lower power,
13
51260
2000
u odnosu na ranije postignute, uz manje energije,
00:53
which enabled the walking biped robots, the first androids.
14
53260
4000
koji je omogućio robotima da hodaju na dve noge, to su prvi androidi.
00:57
So, it's a full range of facial expressions
15
57260
2000
Tako da pričamo o punom spektru izraza lica
00:59
simulating all the major muscles in the human face,
16
59260
2000
koji simuliraju glavne mišiće ljudskog lica,
01:01
running on very small batteries,
17
61260
2000
a napajaju se veoma malim baterijama
01:03
extremely lightweight.
18
63260
2000
neverovatno lake težine.
01:05
The materials that allowed the battery-operated facial expressions
19
65260
3000
Materijal koji je omogućio da uspostavimo ove facijalne ekspresije kojim
01:08
is a material that we call Frubber,
20
68260
2000
upravljaju baterije nazivamo "Frubber",
01:10
and it actually has three major innovations
21
70260
2000
i u njemu su satkana tri glavna izuma
01:12
in the material that allow this to happen.
22
72260
2000
koji omogućavaju da se to desi.
01:14
One is hierarchical pores,
23
74260
2000
Pre svega, hijerarhija pora,
01:16
and the other is a macro-molecular nanoscale porosity in the material.
24
76260
4000
i druga stvar su makro-molekularne nanomreže poroznsoti u materijalu.
01:20
There he's starting to walk.
25
80260
3000
Ovde počinje da korača.
01:23
This is at the Korean Advanced Institute of Science and Technology.
26
83260
3000
Ovo se dešava na Institutu za naprednu nauku i tehnologiju u Koreji.
01:26
I built the head. They built the body.
27
86260
4000
Ja sam napravio glavu. Oni su napravili telo.
01:30
So the goal here is to achieve sentience in machines,
28
90260
3000
Naš cilj je da osposobimo mašine da osećaju
01:33
and not just sentience, but empathy.
29
93260
4000
i ne samo da osećaju, već i da saosećaju.
01:37
We're working with the Machine Perception Laboratory
30
97260
2000
Sarađujemo sa laboratorijom koja se bavi percepcijom kod mašina
01:39
at the U.C. San Diego.
31
99260
2000
na Univerzitetu Kalifornije u San Dijegu.
01:41
They have this really remarkable facial expression technology
32
101260
3000
Oni poseduju zadivljujuću tehnologiju kontrole facijalne ekspresije
01:44
that recognizes facial expressions,
33
104260
2000
pomoću koje je moguće prepoznati izraze lica,
01:46
what facial expressions you're making.
34
106260
2000
tip izraza lica koje ispoljavate.
01:48
It also recognizes where you're looking, your head orientation.
35
108260
3000
Takođe uočava šta posmatrate, položaj vaše glave.
01:51
We're emulating all the major facial expressions,
36
111260
2000
Mi oponašamo glavne izraze lica
01:53
and then controlling it with the software
37
113260
2000
a potom ih uz pomoć softvera kontrolišemo,
01:55
that we call the Character Engine.
38
115260
2000
a to nazivamo "Mašina sa karakterom".
01:57
And here is a little bit of the technology that's involved in that.
39
117260
4000
Ovo je mali prikaz tehnologije koju za to koristimo.
02:01
In fact, right now -- plug it from here, and then plug it in here,
40
121260
8000
U stvari, sada ćemo da ga uključimo ovde, pa uključimo tamo,
02:09
and now let's see if it gets my facial expressions.
41
129260
3000
i videćemo da li uočava moj izraz lica.
02:12
Okay. So I'm smiling.
42
132260
5000
Okej. Smešim se.
02:17
(Laughter)
43
137260
2000
(smeh)
02:19
Now I'm frowning.
44
139260
2000
Sada se mrštim.
02:21
And this is really heavily backlit.
45
141260
4000
Ovo je snažno osvetljeno od pozadi.
02:25
Okay, here we go.
46
145260
2000
Okej, idemo.
02:27
Oh, it's so sad.
47
147260
2000
Ah, tako je tužno.
02:29
Okay, so you smile, frowning.
48
149260
3000
Okej, smešiš se, mrštiš.
02:32
So his perception of your emotional states
49
152260
2000
Njegova percepcija vašeg emotivnog stanja
02:34
is very important for machines to effectively become empathetic.
50
154260
4000
je veoma značajna kako bi mašine postale efektivno empatične.
02:38
Machines are becoming devastatingly capable
51
158260
3000
Mašine postaju nezamislivo vešte u
02:41
of things like killing. Right?
52
161260
4000
ubijanju. Zar ne?
02:45
Those machines have no place for empathy.
53
165260
2000
Kod takvih mašina nema mesta za empatiju.
02:47
And there is billions of dollars being spent on that.
54
167260
2000
Trošimo milijarde dolara na to.
02:49
Character robotics could plant the seed
55
169260
2000
Robotika sa karakterom može utrti put za razvoj
02:51
for robots that actually have empathy.
56
171260
2000
robota koji zaista saosećaju.
02:53
So, if they achieve human level intelligence
57
173260
2000
Ukoliko dosegnu ljudski nivo inteligencije,
02:55
or, quite possibly, greater than human levels of intelligence,
58
175260
4000
ili, što je verovatno, premaše ljudski nivo inteligencije,
02:59
this could be the seeds of hope for our future.
59
179260
3000
to bi moglo da bude seme nade za budućnost.
03:02
So, we've made 20 robots in the last eight years, during the course of getting my Ph.D.
60
182260
4000
U toku poslednjih osam godina, za vreme izrade mog doktorata, napravili smo 20 robota.
03:06
And then I started Hanson Robotics,
61
186260
2000
Potom sam osnovao firmu Henson Robotika,
03:08
which has been developing these things for mass manufacturing.
62
188260
4000
koja se bavi razvojem ovih stvari za masovnu proizvodnju.
03:12
This is one of our robots
63
192260
2000
Ovo je jedan od naših robota kog
03:14
that we showed at Wired NextFest a couple of years ago.
64
194260
2000
smo prikazali na "Wired NextFest" manifestaciji pre dve godine.
03:16
And it sees multiple people in a scene,
65
196260
3000
Ono vidi mnogo ljudi na sceni,
03:19
remembers where individual people are,
66
199260
2000
pamti gde su pojedine osobe,
03:21
and looks from person to person, remembering people.
67
201260
4000
i posmatra i pamti osobu po osobu.
03:25
So, we're involving two things.
68
205260
2000
Uključujemo tu dve stvari.
03:27
One, the perception of people,
69
207260
2000
Pod jedan, uočavanje ljudi
03:29
and two, the natural interface,
70
209260
4000
i pod dva, prirodni izgled uređaja,
03:33
the natural form of the interface,
71
213260
2000
izgled prirodnog oblika,
03:35
so that it's more intuitive for you to interact with the robot.
72
215260
3000
kako bi vam bilo prirodnije da imate interakciju sa robotom.
03:38
You start to believe that it's alive and aware.
73
218260
3000
Počinjete da verujete da je živ i svestan.
03:41
So one of my favorite projects was bringing all this stuff together
74
221260
3000
Jedan od mojih omiljenih projekata je za zadatak imao da sve ovo spoji
03:44
in an artistic display of an android portrait
75
224260
3000
u umetničkom prikazu portreta androida
03:47
of science-fiction writer Philip K. Dick,
76
227260
2000
u liku pisca naučne fantastike Filipa K. Dika,
03:49
who wrote great works like, "Do Androids Dream of Electric Sheep?"
77
229260
3000
on je napisao dela kao što je "Da li androidi sanjaju o električnim ovcama?"
03:52
which was the basis of the movie "Bladerunner."
78
232260
2000
po kom je snimljen film "Bladerunner".
03:54
In these stories, robots often think
79
234260
3000
Često u ovim pričama robot
03:57
that they're human, and they sort of come to life.
80
237260
2000
misli da je ljudsko biće, i na neki način roboti ožive.
03:59
So we put his writings, letters,
81
239260
3000
Tako da smo od njegovih pisanih dela, pisama,
04:02
his interviews, correspondences,
82
242260
3000
intervjua, prepiski,
04:05
into a huge database of thousands of pages,
83
245260
2000
napravili ogormnu bazu podataka od hiljada stranica,
04:07
and then used some natural language processing
84
247260
2000
a potom smo primenili prirodni procesor jezika
04:09
to allow you to actually have a conversation with him.
85
249260
2000
kako bismo vam omogućili da vodite razgovor sa njim.
04:11
And it was kind of spooky, because he would say these things
86
251260
2000
Bilo je zastrašujuće na neki način, jer stvari koje je izgovarao su
04:13
that just sounded like they really understood you.
87
253260
3000
zvučale kao da vas je zaista razumeo.
04:16
And this is one of the most exciting projects that we're developing,
88
256260
3000
Ovo je jedan od najuzbudljivijih projekata koje razvijamo,
04:19
which is a little character that's a spokesbot
89
259260
3000
a to je robot koji priča, mališan
04:22
for friendly artificial intelligence, friendly machine intelligence.
90
262260
3000
koji predstavlja prijateljsku, veštačku inteligenciju, prijateljsku, inteligentnu mašinu.
04:25
And we're getting this mass-manufactured.
91
265260
2000
Mi ovo masovno proizvodimo.
04:27
We specked it out to actually be doable
92
267260
3000
Pojednostavili smo proces tako da možemo to da napravimo
04:30
with a very, very low-cost bill of materials,
93
270260
3000
od veoma jeftinog materijala,
04:33
so that it can become a childhood companion for kids.
94
273260
4000
pa može postati prijatelj u detinjstvu.
04:37
Interfacing with the Internet, it gets smarter over the years.
95
277260
3000
Komunicira sa internetom, postaje pametniji kako godine prolaze.
04:40
As artificial intelligence evolves, so does his intelligence.
96
280260
3000
Njegova inteligencija evoluira sa evolucijom veštačke inteligencije.
04:43
Chris Anderson: Thank you so much. That's incredible.
97
283260
2000
Kris Anderson: Hvala puno. Ovo je neverovatno.
04:45
(Applause)
98
285260
7000
(aplauz)
About this website

This site will introduce you to YouTube videos that are useful for learning English. You will see English lessons taught by top-notch teachers from around the world. Double-click on the English subtitles displayed on each video page to play the video from there. The subtitles scroll in sync with the video playback. If you have any comments or requests, please contact us using this contact form.

https://forms.gle/WvT1wiN1qDtmnspy7