4 lessons from robots about being human
-
0:01 - 0:03I know this is going to sound strange,
-
0:03 - 0:06but I think robots can inspire us
-
0:06 - 0:09to be better humans.
-
0:09 - 0:12See, I grew up in Bethlehem, Pennsylvania,
-
0:12 - 0:15the home of Bethlehem Steel.
-
0:15 - 0:17My father was an engineer,
-
0:17 - 0:20and when I was growing up, he would teach me
-
0:20 - 0:21how things worked.
-
0:21 - 0:24We would build projects together,
-
0:24 - 0:26like model rockets and slot cars.
-
0:26 - 0:30Here's the go-kart that we built together.
-
0:30 - 0:32That's me behind the wheel,
-
0:32 - 0:36with my sister and my best friend at the time,
-
0:36 - 0:38and one day,
-
0:38 - 0:41he came home, when I was about 10 years old,
-
0:41 - 0:43and at the dinner table, he announced
-
0:43 - 0:50that for our next project, we were going to build a robot.
-
0:50 - 0:51A robot.
-
0:51 - 0:53Now, I was thrilled about this,
-
0:53 - 0:55because at school,
-
0:55 - 0:57there was a bully named Kevin,
-
0:57 - 0:59and he was picking on me
-
0:59 - 1:01because I was the only Jewish kid in class.
-
1:01 - 1:04So I couldn't wait to get started to work on this
-
1:04 - 1:08so I could introduce Kevin to my robot. (Laughter)
-
1:08 - 1:19(Robot noises)
-
1:19 - 1:24But that wasn't the kind of robot my dad had in mind.
-
1:24 - 1:28See, he owned a chromium plating company,
-
1:28 - 1:30and they had to move
-
1:30 - 1:33heavy steel parts between tanks of chemicals,
-
1:33 - 1:37and so he needed an industrial robot like this
-
1:37 - 1:40that could basically do the heavy lifting.
-
1:40 - 1:44But my dad didn't get the kind of robot he wanted, either.
-
1:44 - 1:46He and I worked on it for several years,
-
1:46 - 1:48but it was the 1970s,
-
1:48 - 1:51and the technology that was available to amateurs
-
1:51 - 1:53just wasn't there yet.
-
1:53 - 1:57So Dad continued to do this kind of work by hand,
-
1:57 - 2:00and a few years later,
-
2:00 - 2:04he was diagnosed with cancer.
-
2:04 - 2:07You see, what the robot we were trying to build
-
2:07 - 2:10was telling him was not about doing the heavy lifting.
-
2:10 - 2:15It was a warning about his exposure to the toxic chemicals.
-
2:15 - 2:18He didn't recognize that at the time,
-
2:18 - 2:20and he contracted leukemia,
-
2:20 - 2:23and he died at the age of 45.
-
2:23 - 2:26I was devastated by this,
-
2:26 - 2:30and I never forgot the robot that he and I tried to build.
-
2:30 - 2:35When I was in college, I decided to study engineering, like him.
-
2:35 - 2:40And I went to Carnegie Mellon, and I earned my PhD in robotics.
-
2:40 - 2:43I've been studying robots ever since.
-
2:43 - 2:44So what I'd like to tell you about
-
2:44 - 2:47are four robot projects
-
2:47 - 2:54and how they've inspired me to be a better human.
-
2:54 - 3:00By 1993, I was a young professor at USC,
-
3:00 - 3:03and I was just building up my own robotics lab,
-
3:03 - 3:06and this was the year that the World Wide Web came out.
-
3:06 - 3:08And I remember my students were the ones
-
3:08 - 3:09who told me about it,
-
3:09 - 3:12and we would -- we were just amazed.
-
3:12 - 3:15We started playing with this, and that afternoon,
-
3:15 - 3:19we realized that we could use this new, universal interface
-
3:19 - 3:22to allow anyone in the world
-
3:22 - 3:25to operate the robot in our lab.
-
3:25 - 3:30So, rather than have it fight or do industrial work,
-
3:30 - 3:33we decided to build a planter,
-
3:33 - 3:35put the robot into the center of it,
-
3:35 - 3:37and we called it the Telegarden.
-
3:37 - 3:41And we had put a camera in the gripper of the hand
-
3:41 - 3:44of the robot, and we wrote some special scripts
-
3:44 - 3:47and software so that anyone in the world could come in
-
3:47 - 3:49and by clicking on the screen
-
3:49 - 3:51they could move the robot around
-
3:51 - 3:54and visit the garden.
-
3:54 - 3:57But we also allowed, set up some other software
-
3:57 - 4:01that lets you participate and help us water the garden
-
4:01 - 4:04remotely, and if you water it a few times,
-
4:04 - 4:07we'd give you your own seed to plant.
-
4:07 - 4:11Now, this was a project, an engineering project,
-
4:11 - 4:13and we published some papers on the design,
-
4:13 - 4:16the system design of it, but we also thought of it
-
4:16 - 4:19as an art installation.
-
4:19 - 4:21It was invited, after the first year,
-
4:21 - 4:24by the Ars Electronica Museum in Austria
-
4:24 - 4:27to have it installed in their lobby,
-
4:27 - 4:29and I'm happy to say it remained online there,
-
4:29 - 4:3424 hours a day, for almost nine years.
-
4:34 - 4:38That robot was operated by more people
-
4:38 - 4:41than any other robot in history.
-
4:41 - 4:43Now, one day,
-
4:43 - 4:45I got a call out of the blue
-
4:45 - 4:47from a student,
-
4:47 - 4:52who asked a very simple but profound question.
-
4:52 - 4:56He said, "Is the robot real?"
-
4:56 - 4:59Now, everyone else had assumed it was,
-
4:59 - 5:01and we knew it was because we were working with it.
-
5:01 - 5:03But I knew what he meant,
-
5:03 - 5:05because it would be possible to take a bunch of pictures
-
5:05 - 5:10of flowers in a garden and then, basically, index them
-
5:10 - 5:12in a computer system such that it would appear
-
5:12 - 5:15that there was a real robot when there wasn't.
-
5:15 - 5:16And the more I thought about it, I couldn't think
-
5:16 - 5:20of a good answer for how he could tell the difference.
-
5:20 - 5:23This was right about the time that I was offered a position
-
5:23 - 5:25here a Berkeley,
-
5:25 - 5:28and when I got here, I looked up Hubert Dreyfus,
-
5:28 - 5:32who's a world-renowned Professor of Philosophy,
-
5:32 - 5:34and I talked with him about this, and he said,
-
5:34 - 5:38"This is one of the oldest and most central problems
-
5:38 - 5:42in philosophy. It goes back to the Skeptics,
-
5:42 - 5:44and up through Descartes.
-
5:44 - 5:47It's the issue of epistemology,
-
5:47 - 5:51the study of how do we know that something is true."
-
5:51 - 5:53So he and I started working together,
-
5:53 - 5:56and we coined a new term: telepistemology,
-
5:56 - 5:59the study of knowledge at a distance.
-
5:59 - 6:02We invited leading artists, engineers,
-
6:02 - 6:05and philosophers to write essays about this,
-
6:05 - 6:07and the results, the results are collected in this book
-
6:07 - 6:10from MIT Press.
-
6:10 - 6:12So thanks to this student who questioned
-
6:12 - 6:15what everyone else had assumed to be true,
-
6:15 - 6:19this project taught me an important lesson about life,
-
6:19 - 6:23which is to always question assumptions.
-
6:23 - 6:26Now, the second project I'll tell you about
-
6:26 - 6:28grew out of the Telegarden.
-
6:28 - 6:31As it was operating, my students and I were very interested
-
6:31 - 6:33in how people were interacting with each other
-
6:33 - 6:35and what they were doing with the garden.
-
6:35 - 6:37So we started thinking, what if the robot could leave
-
6:37 - 6:39the garden and go out into some other
-
6:39 - 6:41interesting environment?
-
6:41 - 6:43Like, for example, what if it could go to a dinner party
-
6:43 - 6:49at the White House? (Laughter)
-
6:49 - 6:52So, because we were interested more in the system design
-
6:52 - 6:55and the user interface than in the hardware,
-
6:55 - 6:57we decided that, rather than have
-
6:57 - 7:01a robot replace the human to go to the party,
-
7:01 - 7:03we'd have a human replace the robot.
-
7:03 - 7:06We called it the Tele-Actor.
-
7:06 - 7:08We got a human,
-
7:08 - 7:11someone who's very outgoing and gregarious,
-
7:11 - 7:14and she was outfitted with a helmet
-
7:14 - 7:17with various equipment, cameras and microphones,
-
7:17 - 7:20and then a backpack with wireless Internet connection,
-
7:20 - 7:24and the idea was that she could go into a remote and
-
7:24 - 7:28interesting environment, and then over the Internet,
-
7:28 - 7:31people could experience what she was experiencing,
-
7:31 - 7:34so they could see what she was seeing,
-
7:34 - 7:37but then, more importantly, they could participate
-
7:37 - 7:40by interacting with each other
-
7:40 - 7:44and coming up with ideas about what she should do next
-
7:44 - 7:46and where she should go,
-
7:46 - 7:49and then conveying those to the Tele-Actor.
-
7:49 - 7:52So we got a chance to take the Tele-Actor
-
7:52 - 7:55to the Webby Awards in San Francisco,
-
7:55 - 7:59and that year, Sam Donaldson was the host.
-
7:59 - 8:03Just before the curtain went up, I had about 30 seconds
-
8:03 - 8:07to explain to Mr. Donaldson what we were gonna do,
-
8:07 - 8:09and I said, "The Tele-Actor
-
8:09 - 8:12is going to be joining you on stage,
-
8:12 - 8:14and this is a new experimental project,
-
8:14 - 8:16and people are watching her on their screens,
-
8:16 - 8:19and she's got -- there's cameras involved and there's
-
8:19 - 8:22microphones and she's got an earbud in her ear,
-
8:22 - 8:23and people over the network are giving her advice
-
8:23 - 8:25about what to do next."
-
8:25 - 8:28And he said, "Wait a second,
-
8:28 - 8:34that's what I do." (Laughter)
-
8:34 - 8:36So he loved the concept,
-
8:36 - 8:38and when the Tele-Actor walked onstage,
-
8:38 - 8:41she walked right up to him, and she gave him a big kiss
-
8:41 - 8:44right on the lips. (Laughter)
-
8:44 - 8:45We were totally surprised.
-
8:45 - 8:47We had no idea that would happen.
-
8:47 - 8:50And he was great. He just gave her a big hug in return,
-
8:50 - 8:52and it worked out great.
-
8:52 - 8:54But that night, as we were packing up,
-
8:54 - 8:58I asked the Tele-Actor, how did the Tele-Directors
-
8:58 - 9:03decide that they would give a kiss to Sam Donaldson?
-
9:03 - 9:05And she said they hadn't.
-
9:05 - 9:08She said, when she was just about to walk on stage,
-
9:08 - 9:10the Tele-Directors were still trying to agree on what to do,
-
9:10 - 9:12and so she just walked on stage and did
-
9:12 - 9:18what felt most natural. (Laughter)
-
9:18 - 9:22So, the success of the Tele-Actor that night
-
9:22 - 9:26was due to the fact that she was a wonderful actor.
-
9:26 - 9:28She knew when to trust her instincts,
-
9:28 - 9:32and so that project taught me another lesson about life,
-
9:32 - 9:39which is that, when in doubt, improvise. (Laughter)
-
9:39 - 9:42Now, the third project grew out of
-
9:42 - 9:47my experience when my father was in the hospital.
-
9:47 - 9:49He was undergoing a treatment,
-
9:49 - 9:53chemotherapy treatments, and there's a related treatment
-
9:53 - 9:58called brachytherapy, where tiny, radioactive seeds
-
9:58 - 10:02are placed into the body to treat cancerous tumors.
-
10:02 - 10:04And the way it's done, as you can see here,
-
10:04 - 10:08is that surgeons insert needles into the body
-
10:08 - 10:11to deliver the seeds, and all this,
-
10:11 - 10:14all these needles are inserted in parallel,
-
10:14 - 10:17so it's very common that some of the needles
-
10:17 - 10:22penetrate sensitive organs, and as a result,
-
10:22 - 10:27the needles damage these organs, cause damage
-
10:27 - 10:31which leads to trauma and side effects.
-
10:31 - 10:33So my students and I wondered, what if we could
-
10:33 - 10:37modify the system
-
10:37 - 10:40so that the needles could come in at different angles?
-
10:40 - 10:43So we simulated this, and we developed some
-
10:43 - 10:46optimization algorithms and we simulated this,
-
10:46 - 10:48and we were able to show that we are able to avoid
-
10:48 - 10:52the delicate organs and yet still achieve the coverage
-
10:52 - 10:55of the tumors with the radiation.
-
10:55 - 10:59So now, we're working with doctors at UCSF
-
10:59 - 11:02and engineers at Johns Hopkins
-
11:02 - 11:05and we're building a robot that has a number of,
-
11:05 - 11:08it's a specialized design with different joints that can allow
-
11:08 - 11:13the needles to come in at an infinite variety of angles,
-
11:13 - 11:16and as you can see here, they can avoid delicate organs
-
11:16 - 11:20and still reach the targets they're aiming for.
-
11:20 - 11:23So, by questioning this assumption that all the needles
-
11:23 - 11:26have to be parallel, this project also taught me
-
11:26 - 11:29an important lesson: When in doubt --
-
11:29 - 11:34When your path is blocked, pivot.
-
11:34 - 11:38And the last project also has to do with medical robotics.
-
11:38 - 11:42And this is something that's grown out of a system called
-
11:42 - 11:46the da Vinci surgical robot,
-
11:46 - 11:48and this is a commercially available device.
-
11:48 - 11:52It's being used in over 2,000 hospitals around the world,
-
11:52 - 11:54and the idea is it allows the surgeon
-
11:54 - 11:58to operate comfortably in his own coordinate frame,
-
11:58 - 12:03but many of the subtasks in surgery
-
12:03 - 12:06are very routine and tedious, like suturing,
-
12:06 - 12:09and currently, all of these are performed
-
12:09 - 12:13under the specific and immediate control of the surgeon,
-
12:13 - 12:16so the surgeon becomes fatigued over time.
-
12:16 - 12:17And we've been wondering,
-
12:17 - 12:19what if we could program the robot
-
12:19 - 12:22to perform some of these subtasks,
-
12:22 - 12:24and thereby free the surgeons to focus
-
12:24 - 12:26on the more complicated parts of the surgery,
-
12:26 - 12:30and also cut down on the time that the surgery would take
-
12:30 - 12:33if we could get the robot to do them a little bit faster?
-
12:33 - 12:35Now, it's hard to program a robot to do delicate things
-
12:35 - 12:39like this, but it turns out my colleague, Pieter Abbeel,
-
12:39 - 12:42who's here at Berkeley, has develeloped
-
12:42 - 12:47a new set of techniques for teaching robots from example.
-
12:47 - 12:50So he's gotten robots to fly helicopters,
-
12:50 - 12:53do incredibly interesting, beautiful acrobatics,
-
12:53 - 12:56by watching human experts fly them.
-
12:56 - 12:58So we got one of these robots.
-
12:58 - 13:01We started working with Pieter and his students,
-
13:01 - 13:03and we asked a surgeon to perform
-
13:03 - 13:08a task, and what we do is we, with the robot,
-
13:08 - 13:10so what we're doing is asking the robot,
-
13:10 - 13:11the surgeon to perform the task,
-
13:11 - 13:13and we record the motions of the robot.
-
13:13 - 13:15So here's an example. I'll use a figure eight,
-
13:15 - 13:18tracing out a figure eight as an example.
-
13:18 - 13:21So here's what it looks like when the robot,
-
13:21 - 13:24this is what the robot's path looks like,
-
13:24 - 13:25those three examples.
-
13:25 - 13:27Now, those are much better than what a novice
-
13:27 - 13:32like I could do, but they're still jerky and imprecise.
-
13:32 - 13:34So we record all these examples, the data,
-
13:34 - 13:38and then we go through a sequence of steps.
-
13:38 - 13:41First, we used a technique called dynamic time warping
-
13:41 - 13:43from speech recognition, and this allows us to
-
13:43 - 13:46temporally align all of the examples,
-
13:46 - 13:49and then we apply Kalman filtering,
-
13:49 - 13:52a technique from control theory, that allows us
-
13:52 - 13:55to statistically analyze all the noise
-
13:55 - 14:01and extract the desired trajectory that underlies them.
-
14:01 - 14:03Now, so what we're doing is that we take those
-
14:03 - 14:05human demonstrations, they're all noisy and imperfect,
-
14:05 - 14:08and we extract from them an inferred task trajectory
-
14:08 - 14:11and control sequence for the robot.
-
14:11 - 14:13We then execute that on the robot,
-
14:13 - 14:16we observe what happens,
-
14:16 - 14:18then we adjust the controls using a sequence of techniques
-
14:18 - 14:21called iterative learning.
-
14:21 - 14:25Then what we do is, we increase the velocity a little bit.
-
14:25 - 14:29We observe the results, adjust the controls again,
-
14:29 - 14:31and observe what happens.
-
14:31 - 14:33And we go through this several rounds.
-
14:33 - 14:35And here's the result.
-
14:35 - 14:37That's the inferred task trajectory,
-
14:37 - 14:40and here's the robot moving at the speed of the human.
-
14:40 - 14:42Here's four times the speed of the human.
-
14:42 - 14:45Here's seven times.
-
14:45 - 14:49And here's the robot operating at 10 times
-
14:49 - 14:51the speed of the human.
-
14:51 - 14:54So we're able to get a robot to perform a delicate task,
-
14:54 - 14:57like a surgical subtask,
-
14:57 - 15:00at 10 times the speed of a human.
-
15:00 - 15:04So this project also, because of its involved practicing
-
15:04 - 15:07and learning, doing something over and over again,
-
15:07 - 15:09this project also has a lesson, which is,
-
15:09 - 15:13if you want to do something well,
-
15:13 - 15:20there's no substitute for practice, practice, practice.
-
15:21 - 15:24So these are four of the lessons that I've learned
-
15:24 - 15:27from robots over the years,
-
15:27 - 15:32and robotics, the field of robotics has gotten much better
-
15:32 - 15:34over time.
-
15:34 - 15:36Nowadays, high school students can build robots
-
15:36 - 15:40like the industrial robot my dad and I tried to build.
-
15:40 - 15:47And now, I have a daughter,
-
15:47 - 15:50named Odessa.
-
15:50 - 15:52She's eight years old,
-
15:52 - 15:54and she likes robots, too.
-
15:54 - 15:57Maybe it runs in the family. (Laughter)
-
15:57 - 16:00I wish she could meet my dad.
-
16:00 - 16:03And now I get to teach her how things work,
-
16:03 - 16:06and we get to build projects together, and I wonder
-
16:06 - 16:10what kind of lessons that she'll learn from them.
-
16:10 - 16:13Robots are the most human
-
16:13 - 16:15of our machines.
-
16:15 - 16:18They can't solve all of the world's problems,
-
16:18 - 16:22but I think they have something important to teach us.
-
16:22 - 16:26I invite all of you to think about the innovations
-
16:26 - 16:28that you're interested in,
-
16:28 - 16:32the machines that you wish for,
-
16:32 - 16:35and think about what they might be telling you,
-
16:35 - 16:37because I have a hunch
-
16:37 - 16:39that many of our technological innovations,
-
16:39 - 16:42the devices we dream about,
-
16:42 - 16:46can inspire us to be better humans.
-
16:46 - 16:49Thank you. (Applause)
- Title:
- 4 lessons from robots about being human
- Speaker:
- Ken Goldberg
- 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. (Filmed at TEDxBerkeley.)
- Video Language:
- English
- Team:
- closed TED
- Project:
- TEDTalks
- Duration:
- 17:09
Krystian Aparta commented on English subtitles for 4 lessons from robots about being human | ||
Krystian Aparta edited English subtitles for 4 lessons from robots about being human | ||
Krystian Aparta edited English subtitles for 4 lessons from robots about being human | ||
Krystian Aparta edited English subtitles for 4 lessons from robots about being human | ||
Morton Bast edited English subtitles for 4 lessons from robots about being human | ||
Morton Bast edited English subtitles for 4 lessons from robots about being human | ||
Thu-Huong Ha approved English subtitles for 4 lessons from robots about being human | ||
Thu-Huong Ha accepted English subtitles for 4 lessons from robots about being human |
Krystian Aparta
The English transcript was updated on 5/9/2016.