4 lessons from robots about being human | Ken Goldberg | TEDxBerkeley
-
0:30 - 0:34Thank you. It's been an amazing
line-up of speakers today. -
0:34 - 0:37We've been talking backstage
and I just want to say -
0:37 - 0:41we've all agreed that you have been
a terrific audience. -
0:41 - 0:44So I think you all deserve
a round of applause for being so great. -
0:44 - 0:48(Applause)
-
0:50 - 0:53I know this is going to sound strange,
-
0:53 - 0:58but I think robots can inspire us
to be better humans. -
0:59 - 1:02See, I grew up in Bethlehem, Pennsylvania,
-
1:02 - 1:04the home of Bethlehem Steel.
-
1:04 - 1:07My father was an engineer,
-
1:07 - 1:11and when I was growing up,
he would teach me how things worked. -
1:11 - 1:13We would build projects together,
-
1:13 - 1:16like model rockets and slot cars.
-
1:16 - 1:19Here's the go-kart that we built together.
-
1:20 - 1:21That's me behind the wheel,
-
1:21 - 1:24with my sister and my best
friend at the time. -
1:25 - 1:27And one day,
-
1:27 - 1:30he came home, when I was
about 10 years old, -
1:30 - 1:33and at the dinner table, he announced
-
1:33 - 1:37that for our next project,
we were going to build ... -
1:37 - 1:38a robot.
-
1:39 - 1:40A robot.
-
1:41 - 1:43Now, I was thrilled about this,
-
1:43 - 1:47because at school,
there was a bully named Kevin, -
1:47 - 1:48and he was picking on me,
-
1:48 - 1:51because I was the only
Jewish kid in class. -
1:51 - 1:54So I couldn't wait to get
started to work on this, -
1:54 - 1:56so I could introduce Kevin to my robot.
-
1:56 - 1:57(Laughter)
-
1:57 - 2:02(Robot noises)
-
2:07 - 2:08(Laughter)
-
2:08 - 2:12But that wasn't the kind of robot
my dad had in mind. -
2:12 - 2:13(Laughter)
-
2:13 - 2:17See, he owned a chromium-plating company,
-
2:17 - 2:23and they had to move heavy steel parts
between tanks of chemicals. -
2:23 - 2:26And so he needed
an industrial robot like this, -
2:26 - 2:28that could basically do the heavy lifting.
-
2:29 - 2:33But my dad didn't get
the kind of robot he wanted, either. -
2:33 - 2:36He and I worked on it for several years,
-
2:36 - 2:41but it was the 1970s, and the technology
that was available to amateurs -
2:41 - 2:42just wasn't there yet.
-
2:43 - 2:46So Dad continued to do
this kind of work by hand. -
2:47 - 2:49And a few years later,
-
2:49 - 2:51he was diagnosed with cancer.
-
2:53 - 2:55You see,
-
2:55 - 2:58what the robot we were trying
to build was telling him -
2:58 - 3:00was not about doing the heavy lifting.
-
3:00 - 3:01It was a warning
-
3:01 - 3:03about his exposure to the toxic chemicals.
-
3:05 - 3:07He didn't recognize that at the time,
-
3:07 - 3:09and he contracted leukemia.
-
3:09 - 3:11And he died at the age of 45.
-
3:13 - 3:15I was devastated by this.
-
3:15 - 3:18And I never forgot the robot
that he and I tried to build. -
3:20 - 3:23When I was at college, I decided
to study engineering, like him. -
3:25 - 3:29And I went to Carnegie Mellon,
and I earned my PhD in robotics. -
3:29 - 3:31I've been studying robots ever since.
-
3:32 - 3:37So what I'd like to tell you about
are four robot projects, -
3:37 - 3:40and how they've inspired me
to be a better human. -
3:44 - 3:49By 1993, I was a young professor at USC,
-
3:49 - 3:52and I was just building up
my own robotics lab, -
3:52 - 3:55and this was the year
the World Wide Web came out. -
3:56 - 3:59And I remember my students
were the ones who told me about it, -
3:59 - 4:01and we would -- we were just amazed.
-
4:01 - 4:05We started playing with this,
and that afternoon, -
4:05 - 4:09we realized that we could use
this new, universal interface -
4:09 - 4:13to allow anyone in the world
to operate the robot in our lab. -
4:15 - 4:19So, rather than have it fight
or do industrial work, -
4:20 - 4:22we decided to build a planter,
-
4:22 - 4:24put the robot into the center of it,
-
4:24 - 4:26and we called it the Telegarden.
-
4:27 - 4:32And we had put a camera
in the gripper of the hand of the robot, -
4:32 - 4:35and we wrote some
special scripts and software, -
4:35 - 4:37so that anyone in the world could come in,
-
4:37 - 4:39and by clicking on the screen,
-
4:39 - 4:42they could move the robot around
and visit the garden. -
4:43 - 4:47But we also set up some other software
-
4:47 - 4:51that lets you participate
and help us water the garden, remotely. -
4:51 - 4:53And if you watered it a few times,
-
4:53 - 4:56we'd give you your own seed to plant.
-
4:57 - 5:00Now, this was an engineering project,
-
5:00 - 5:04and we published some papers
on the system design of it, -
5:04 - 5:07but we also thought of it
as an art installation. -
5:08 - 5:10It was invited, after the first year,
-
5:10 - 5:13by the Ars Electronica Museum in Austria,
-
5:13 - 5:16to have it installed in their lobby.
-
5:17 - 5:21And I'm happy to say, it remained
online there, 24 hours a day, -
5:21 - 5:23for almost nine years.
-
5:24 - 5:28That robot was operated by more people
-
5:28 - 5:30than any other robot in history.
-
5:31 - 5:32Now, one day,
-
5:32 - 5:36I got a call out of the blue
from a student, -
5:37 - 5:41who asked a very simple
but profound question. -
5:42 - 5:45He said, "Is the robot real?"
-
5:46 - 5:48Now, everyone else had assumed it was,
-
5:48 - 5:51and we knew it was,
because we were working with it. -
5:51 - 5:52But I knew what he meant,
-
5:52 - 5:54because it would be possible
-
5:54 - 5:57to take a bunch of pictures
of flowers in a garden -
5:57 - 6:00and then, basically, index them
in a computer system, -
6:00 - 6:03such that it would appear
that there was a real robot, -
6:03 - 6:04when there wasn't.
-
6:05 - 6:06And the more I thought about it,
-
6:06 - 6:10I couldn't think of a good answer
for how he could tell the difference. -
6:10 - 6:13This was right about the time
that I was offered a position -
6:13 - 6:14here at Berkeley.
-
6:14 - 6:16And when I got here,
-
6:16 - 6:18I looked up Hubert Dreyfus,
-
6:18 - 6:21who's a world-renowned
professor of philosophy, -
6:22 - 6:25And I talked with him
about this and he said, -
6:25 - 6:29"This is one of the oldest
and most central problems in philosophy. -
6:29 - 6:33It goes back to the Skeptics
and up through Descartes. -
6:33 - 6:36It's the issue of epistemology,
-
6:36 - 6:39the study of how do we know
that something is true." -
6:40 - 6:42So he and I started working together,
-
6:42 - 6:45and we coined a new term:
"telepistemology," -
6:46 - 6:48the study of knowledge at a distance.
-
6:49 - 6:53We invited leading artists,
engineers and philosophers -
6:53 - 6:55to write essays about this,
-
6:55 - 6:58and the results are collected
in this book from MIT Press. -
6:59 - 7:02So thanks to this student,
-
7:02 - 7:05who questioned what everyone else
had assumed to be true, -
7:05 - 7:09this project taught me
an important lesson about life, -
7:09 - 7:12which is to always question assumptions.
-
7:13 - 7:15Now, the second project
I'll tell you about -
7:16 - 7:17grew out of the Telegarden.
-
7:18 - 7:20As it was operating, my students
and I were very interested -
7:20 - 7:23in how people were interacting
with each other, -
7:23 - 7:25and what they were doing with the garden.
-
7:25 - 7:26So we started thinking:
-
7:26 - 7:28what if the robot could leave the garden
-
7:28 - 7:31and go out into some other
interesting environment? -
7:31 - 7:34Like, for example,
what if it could go to a dinner party -
7:34 - 7:35at the White House?
-
7:35 - 7:37(Laughter)
-
7:38 - 7:41So, because we were interested
more in the system design -
7:41 - 7:44and the user interface
than in the hardware, -
7:44 - 7:46we decided that,
-
7:46 - 7:50rather than have a robot replace
the human to go to the party, -
7:50 - 7:52we'd have a human replace the robot.
-
7:53 - 7:55We called it the Tele-Actor.
-
7:55 - 7:57We got a human,
-
7:58 - 8:01someone who's very
outgoing and gregarious, -
8:01 - 8:05and she was outfitted with a helmet
with various equipment, -
8:05 - 8:06cameras and microphones,
-
8:06 - 8:09and then a backpack with wireless
Internet connection. -
8:10 - 8:13And the idea was that she could go
-
8:13 - 8:15into a remote and interesting environment,
-
8:15 - 8:17and then over the Internet,
-
8:17 - 8:20people could experience
what she was experiencing. -
8:20 - 8:23So they could see what she was seeing,
-
8:23 - 8:27but then, more importantly,
they could participate, -
8:27 - 8:31by interacting with each other
and coming up with ideas -
8:31 - 8:35about what she should do next
and where she should go, -
8:35 - 8:38and then conveying those
to the Tele-Actor. -
8:39 - 8:41So we got a chance to take the Tele-Actor
-
8:41 - 8:44to the Webby Awards in San Francisco.
-
8:45 - 8:48And that year, Sam Donaldson was the host.
-
8:50 - 8:52Just before the curtain went
up, I had about 30 seconds -
8:53 - 8:56to explain to Mr. Donaldson
what we were going to do. -
8:58 - 9:01And I said, "The Tele-Actor
is going to be joining you onstage. -
9:01 - 9:04This is a new experimental project,
-
9:04 - 9:06and people are watching her
on their screens, -
9:06 - 9:09there's cameras involved
and there's microphones -
9:09 - 9:11and she's got an earbud in her ear,
-
9:11 - 9:14and people over the network
are giving her advice -
9:14 - 9:15about what to do next."
-
9:15 - 9:16And he said, "Wait a second.
-
9:17 - 9:18That's what I do."
-
9:19 - 9:23(Laughter)
-
9:24 - 9:25So he loved the concept,
-
9:25 - 9:30and when the Tele-Actor walked onstage,
she walked right up to him, -
9:30 - 9:32and she gave him a big kiss
right on the lips. -
9:32 - 9:34(Laughter)
-
9:34 - 9:37We were totally surprised --
we had no idea that would happen. -
9:37 - 9:40And he was great, he just gave her
a big hug in return, -
9:40 - 9:41and it worked out great.
-
9:41 - 9:43But that night, as we were packing up,
-
9:43 - 9:48I asked the Tele-Actor,
how did the Tele-Directors decide -
9:48 - 9:51that they would give
a kiss to Sam Donaldson? -
9:53 - 9:54And she said they hadn't.
-
9:55 - 9:57She said, when she was
just about to walk onstage, -
9:57 - 10:00the Tele-Directors still were trying
to agree on what to do, -
10:00 - 10:04and so she just walked onstage
and did what felt most natural. -
10:04 - 10:08(Laughter)
-
10:08 - 10:11So, the success
of the Tele-Actor that night -
10:11 - 10:15was due to the fact
that she was a wonderful actor. -
10:15 - 10:18She knew when to trust her instincts.
-
10:18 - 10:22And so that project taught me
another lesson about life, -
10:22 - 10:26which is that, when in doubt, improvise.
-
10:26 - 10:27(Laughter)
-
10:28 - 10:33Now, the third project
grew out of my experience -
10:33 - 10:35when my father was in the hospital.
-
10:37 - 10:40He was undergoing a treatment --
chemotherapy treatments -- -
10:40 - 10:44and there's a related treatment
called brachytherapy, -
10:44 - 10:48where tiny, radioactive seeds
are placed into the body -
10:48 - 10:50to treat cancerous tumors.
-
10:51 - 10:53And the way it's done,
as you can see here, -
10:53 - 10:58is that surgeons
insert needles into the body -
10:58 - 10:59to deliver the seeds.
-
10:59 - 11:03And all these needles
are inserted in parallel. -
11:04 - 11:09So it's very common that some
of the needles penetrate sensitive organs. -
11:10 - 11:17And as a result, the needles damage
these organs, cause damage, -
11:17 - 11:19which leads to trauma and side effects.
-
11:20 - 11:22So my students and I wondered:
-
11:22 - 11:26what if we could modify the system,
-
11:26 - 11:29so that the needles
could come in at different angles? -
11:30 - 11:32So we simulated this;
-
11:32 - 11:36we developed some optimization
algorithms and we simulated this. -
11:36 - 11:37And we were able to show
-
11:37 - 11:39that we are able to avoid
the delicate organs, -
11:39 - 11:44and yet still achieve the coverage
of the tumors with the radiation. -
11:45 - 11:48So now, we're working with doctors at UCSF
-
11:48 - 11:51and engineers at Johns Hopkins,
-
11:51 - 11:55and we're building a robot
that has a number of -- -
11:55 - 11:57it's a specialized design
with different joints -
11:57 - 12:02that can allow the needles to come in
at an infinite variety of angles. -
12:02 - 12:06And as you can see here,
they can avoid delicate organs -
12:06 - 12:08and still reach the targets
they're aiming for. -
12:10 - 12:15So, by questioning this assumption
that all the needles have to be parallel, -
12:15 - 12:17this project also taught me
an important lesson: -
12:18 - 12:22When in doubt, when your path
is blocked, pivot. -
12:23 - 12:27And the last project
also has to do with medical robotics. -
12:28 - 12:31And this is something
that's grown out of a system -
12:31 - 12:35called the da Vinci surgical robot.
-
12:35 - 12:38And this is a commercially
available device. -
12:38 - 12:41It's being used in over 2,000
hospitals around the world. -
12:42 - 12:46The idea is it allows the surgeon
to operate comfortably -
12:46 - 12:48in his own coordinate frame.
-
12:50 - 12:56Many of the subtasks in surgery are very
routine and tedious, like suturing, -
12:56 - 12:58and currently, all of these are performed
-
12:58 - 13:02under the specific and immediate
control of the surgeon. -
13:03 - 13:05So the surgeon becomes fatigued over time.
-
13:06 - 13:07And we've been wondering,
-
13:07 - 13:11what if we could program the robot
to perform some of these subtasks, -
13:11 - 13:13and thereby free the surgeon
-
13:13 - 13:16to focus on the more complicated
parts of the surgery, -
13:16 - 13:19and also cut down on the time
that the surgery would take -
13:19 - 13:22if we could get the robot
to do them a little bit faster? -
13:22 - 13:26Now, it's hard to program a robot
to do delicate things like this. -
13:26 - 13:31But it turns out my colleague
Pieter Abbeel, who's here at Berkeley, -
13:31 - 13:36has developed a new set of techniques
for teaching robots from example. -
13:37 - 13:39So he's gotten robots to fly helicopters,
-
13:39 - 13:43do incredibly interesting,
beautiful acrobatics, -
13:43 - 13:45by watching human experts fly them.
-
13:46 - 13:48So we got one of these robots.
-
13:48 - 13:50We started working with Pieter
and his students. -
13:50 - 13:54And we asked a surgeon
to perform a task -- -
13:56 - 13:57with the robot.
-
13:57 - 14:00So what we're doing is asking
the surgeon to perform the task, -
14:00 - 14:03and we record the motions of the robot.
-
14:03 - 14:04So here's an example.
-
14:04 - 14:07I'll use tracing out
a figure eight as an example. -
14:08 - 14:11So here's what it looks like
when the robot -- -
14:11 - 14:14this is what the robot's path
looks like, those three examples. -
14:14 - 14:19Now, those are much better
than what a novice like me could do, -
14:19 - 14:21but they're still jerky and imprecise.
-
14:21 - 14:24So we record all these examples, the data,
-
14:24 - 14:27and then go through a sequence of steps.
-
14:28 - 14:31First, we use a technique
called dynamic time warping -
14:31 - 14:32from speech recognition.
-
14:32 - 14:36And this allows us to temporally
align all of the examples. -
14:36 - 14:42And then we apply Kalman filtering,
a technique from control theory, -
14:42 - 14:45that allows us to statistically
analyze all the noise -
14:45 - 14:49and extract the desired
trajectory that underlies them. -
14:51 - 14:53Now we take those human demonstrations --
-
14:53 - 14:55they're all noisy and imperfect --
-
14:55 - 14:58and we extract from them
an inferred task trajectory -
14:58 - 15:00and control sequence for the robot.
-
15:01 - 15:03We then execute that on the robot,
-
15:03 - 15:05we observe what happens,
-
15:05 - 15:06then we adjust the controls,
-
15:06 - 15:09using a sequence of techniques
called iterative learning. -
15:11 - 15:14Then what we do is we increase
the velocity a little bit. -
15:15 - 15:18We observe the results,
adjust the controls again, -
15:19 - 15:21and observe what happens.
-
15:21 - 15:23And we go through this several rounds.
-
15:23 - 15:24And here's the result.
-
15:24 - 15:26That's the inferred task trajectory,
-
15:26 - 15:29and here's the robot
moving at the speed of the human. -
15:29 - 15:32Here's four times the speed of the human.
-
15:32 - 15:33Here's seven times.
-
15:35 - 15:39And here's the robot operating
at 10 times the speed of the human. -
15:40 - 15:43So we're able to get a robot
to perform a delicate task -
15:43 - 15:46like a surgical subtask,
-
15:47 - 15:48at 10 times the speed of a human.
-
15:50 - 15:52So this project also,
-
15:52 - 15:54because of its involved
practicing and learning, -
15:55 - 15:56doing something over and over again,
-
15:56 - 15:59this project also has a lesson, which is:
-
15:59 - 16:01if you want to do something well,
-
16:03 - 16:07there's no substitute
for practice, practice, practice. -
16:11 - 16:14So these are four of the lessons
that I've learned from robots -
16:14 - 16:15over the years.
-
16:17 - 16:22And the field of robotics
has gotten much better over time. -
16:24 - 16:26Nowadays, high school students
can build robots, -
16:26 - 16:29like the industrial robot
my dad and I tried to build. -
16:30 - 16:33But, it's very -- now ...
-
16:33 - 16:37And now, I have a daughter,
-
16:37 - 16:39named Odessa.
-
16:39 - 16:40She's eight years old.
-
16:41 - 16:43And she likes robots, too.
-
16:43 - 16:45Maybe it runs in the family.
-
16:45 - 16:46(Laughter)
-
16:46 - 16:48I wish she could meet my dad.
-
16:50 - 16:53And now I get to teach her
how things work, -
16:53 - 16:55and we get to build projects together.
-
16:55 - 16:58And I wonder what kind of lessons
she'll learn from them. -
17:00 - 17:04Robots are the most human of our machines.
-
17:04 - 17:07They can't solve all
of the world's problems, -
17:07 - 17:11but I think they have something
important to teach us. -
17:12 - 17:14I invite all of you
-
17:14 - 17:17to think about the innovations
that you're interested in, -
17:18 - 17:20the machines that you wish for.
-
17:21 - 17:24And think about
what they might be telling you. -
17:25 - 17:28Because I have a hunch that many
of our technological innovations, -
17:28 - 17:30the devices we dream about,
-
17:32 - 17:35can inspire us to be better humans.
-
17:35 - 17:37Thank you.
-
17:37 - 17:39(Applause)
- Title:
- 4 lessons from robots about being human | Ken Goldberg | TEDxBerkeley
- Description:
-
The more that robots ingrain themselves into our everyday lives, the more we're forced to examine ourselves as people. At TEDxBerkeley, Ken Goldberg shares four very human lessons that he's learned from working with robots.
This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at http://ted.com/tedx
- Video Language:
- English
- Team:
- closed TED
- Project:
- TEDxTalks
- Duration:
- 17:44
TED Translators admin edited English subtitles for TEDxBerkeley - Ken Goldberg - Can Robots Inspire Us To Be Better Humans? | ||
TED Translators admin edited English subtitles for TEDxBerkeley - Ken Goldberg - Can Robots Inspire Us To Be Better Humans? |