200th Episode Special, with Rodney Brooks
interview by Robots Podcast
In this episode, we celebrate our 200th episode! That’s over 6000 minutes of robot goodness and nearly 8 years releasing interviews with your favorite roboticists. The podcast is all volunteer run, a special thanks to everyone on the team who’s made this possible! And thanks to all of you for listening in all these years.
To celebrate, our president Audrow Nash has invited a team of old-timers from the podcast team, and one of our favorite recurring guests, Rodney Brooks from Rethink Robotics.
The panel is composed of Sabine Hauert, Markus Waibel, and Peter Dürr, all co-founders of the podcast, and Per Sjöborg who’s been part of the team for many years now. They discuss their favorite episodes, what area in robotics is most interesting, their best memories, and lessons learned.
Rodney Brooks tells us about latest developments at Rethink Robotics, including their new robot Sawyer, and the Baxter Research Robot. He also tells us about the importance of science communication in a field that is prone to hype and disinformation.
A mathematics undergraduate in his native Australia, Rodney received a Ph.D. in Computer Science from Stanford in 1981. From 1984 to 2010, he was on the MIT faculty, and completed his service as a Professor of Robotics. He was also the founding Director of the Institute’s Computer Science and Artificial Intelligence Laboratory, and served in that role until 2007. In 1990, he co-founded iRobot (NASDAQ: IRBT), where he served variously as CTO, Chairman and board member until 2011. Rodney has been honored by election to the National Academy of Engineering, and has been elected as a Fellow of the American Academy of Arts and Sciences, the Association of Computing Machinery, the Association for the Advancement of Artificial Intelligence, the Institute of Electrical and Electronics Engineers and the American Association for the Advancement of Science. He is the now the founder, chairman, and CTO of Rethink Robotics.
- Download mp3 (17.2 MB)
- Subscribe to Robots using iTunes
- Subscribe to Robots using RSS
- Rethink Robotics’ website
This transcript has been edited for clarity.
Part One: Panel discussion with the Robots Podcast Team
Audrow: Our first question to the panel is: “What have been your favorite robots podcast episodes?”
Sabine: That was a really tough question. There have been so many episodes over the years – 200 of them – so it is pretty hard to choose. Dario Floreano was one of them; he’s one of the reasons the Robots Podcast exists and he started this all out of EPFL. I also really liked interviewing Bob Full about bio-inspired robots. I also interviewed Rodney Brooks a couple of times, including at the offices of Rethink Robotics. That was fun because I got to tour Rethink Robotics and have a chat with Rod over in his office.
Then there were things that were really outside the scope of what I was used to discussing or studying, like the interview with Ryan Calo, which was about the legal aspects of robotics. That’s something I hadn’t really thought about before, and talking to Ryan back then was quite helpful for me in terms of framing that discussion.
And also Radhika Nagpal – just because she’s in an area that I love and she’s a great researcher.
Audrow: What about you, Peter?
Peter: I’m with Sabine: it’s been really difficult to choose. As the audio editor for the podcast, I get to listen to all the episodes before anyone else and I’m always looking forward to each and every new episode. If you had to pin me down on one, I’m going to go for #166: Quest for Computer Vision, with Peter Corke. I work with computer vision in my day job and I really liked Peter Corke’s insightful interview and his overview over this super interesting field.
Sabine: That interview was one you did, Audrow!
Audrow: Yes, it is – thank you! … Per, what has been your favorite episode?
Per: I like the episode from the precursor to the Robots Podcast – Talking Robots – with Robin Murphy from Texas A&M, about rescue robotics. Designing robots that actually work and do the task they’re assigned to in a very hard environment – that was great content.
The favorite interview I did had to be the with Mark Tilden in Hong Kong in his amazing apartment. There were probably – I kid you not – 5000 robots there … everything from the Wowwee robots to his BEAM Robotics stuff. It was just amazing to hear his enormous experience and see all those robots, so that was the favorite one to record for me.
But as everyone else has said, there are so many good interviews and I’ve been fortunate enough to meet so many great people. It’s really impossible to choose.
Audrow: What about you, Markus?
Markus: I’ve not been able to pick a single episode either, but I have really enjoyed the business related episodes. I guess that’s how my personal interests have evolved over the years. When we started back in 2006 with Talking Robots, our first 40 or 45 interviews were almost entirely with academics. That was about 10 years ago, and that was what was out there, mostly. But if you look over the interviews we’ve done in the past year, it turns out to be about half industry and half academia, and I think that’s an interesting development for the field.
Sabine: What’s your favorite, Audrow?
Audrow: Like everyone has been saying, it’s really hard to pick. But I think one of my favorites was # 98 on self-organizing systems with Radhika. That episode gave a great overview of the field and it was interesting listening to her describe it. Also #101 on the history and outlook of robotics with George Bekey and Rodney Brooks, because it gave a big picture of what they had seen in field. It was interesting having the interviewer pick into their wisdom, to give me a better perspective on robotics. Those have been my favorite interviews so far.
My next question for everyone is what area of robotics do you think is the most interesting, and why?
Markus: I work with drones every day – small flying machines – so obviously I have a huge interest there.
Apart from that, one area I’ve been really very excited about from the very start is the field of exoskeletons and prosthetics. I think there’s so much potential to increase our independence as we age with these technologies, and as a hobby, I closely follow how these technologies are developing and being brought to market.
Another one is speech recognition. It’s really more AI [than robotics], but I think speech recognition will completely revolutionize the way we interact with technology. Right now data is mostly entered by keyboard or by mouse, and we can’t really communicate with machines in a natural way. I think speak recognition will revolutionize the way we interact with technology across the board if we manage to solve the privacy issues associated with it – which will not be easy.
Of course the next step is then doing something with that speech. I think the prevalent solution right now is to do some type of cloud processing. But with robots like Jibo or Amazon’s Echo, there’s always the big question of whether I really want Amazon or whoever else to listen in on all the conversations I’m having in my house or in my car or in my office. I think it will be difficult to role this kind of technology out in a major way before the privacy issue is resolved.
Audrow: What about you, Per?
Per: Everybody who knows me knows that I love modular robotics and self-reconfiguring modular robotics. There are so many modular systems coming out now (even if they are not self-reconfiguring) – practical systems that are available commercially … off the shelf … and that’s really exciting to me.
I agree with Markus that we’re seeing more startups, and more business being generated around robotics, and also value provided to the customer. That’s probably one of the most exciting things I see in field of robotics as a whole.
Audrow: Peter, tell us what is most interesting to you and why?
Peter: I’m really excited about artificial intelligence. The field has made huge progress over the last few decades. We’ve been able to create computers that can beat humans at chess, and that can not only recognize images (dogs, for example), but can differentiate between them (i.e. different kinds of dogs). They can surpass humans at recognizing human faces, and so on.
We cannot match the human brain (or even animal brains) at certain tasks, but I think we’ve made huge progress in artificial intelligence and I cannot wait to see how intelligent robots will be in the near future. That’s what really excites me.
Audrow: Sabine, what is most interesting and why?
Sabine: Ten years ago when we were starting the podcast, we were worried we wouldn’t have enough material to keep the podcast going every two weeks … that we’d somehow run out of topics. That definitely hasn’t happened. Looking at the topics from the past 200 episodes, they are all very different – it really shows the breadth of the field.
If I had to pick one (and I’m biased because this is the field I work in), I’m quite excited about swarm robotics. The reason is because we have finally reached a point where we can make a thousand plus robots and have them coordinate, and these are the sheer numbers that start to make swarm robotics interesting.
An obvious application is the biomedical field because, whether you’re talking about cells or nano-robots or molecules or chemistry, these are things that work in the trillions. In large swarm numbers, a very simple entity could be helpful for detecting cancers or doing something interesting in the body. That’s the field I’m excited about right now.
Audrow: What are your best memories with the Robots Podcast, Sabine?
Sabine: Starting up the whole podcast, and getting it going.
Markus was doing the episodes for Talking Robots, along with Peter, who has always been there doing the audio. We were all at Dario Floreano’s lab at EPFL in those days, and we used to move the couch against the wall to help soundproof the room so there wouldn’t be any echo. Peter is a professional musician, so he had this super sophisticated setup with real microphones, like the ones used in radio shows. We’d have this whole setup right in the laboratory. It was really fun.
In the beginning it took a lot of time. Audrow, you’ve done an amazing job in terms of continuing the podcast and making sure that we can keep going by having it run so nicely. What’s also fun is to see how it’s evolved into something that works, that’s efficient, and to see just how global it’s become. Per is in Tokyo, Markus is in Switzerland, I’m based in the UK, and Audrow, you’re in the US. Seeing that evolution from all of us in this office at EPFL, up to something that’s this global, that’s been great.
Peter: Now that Sabine is telling the anecdotes from the early days, I have to agree that was certainly fun. I think for me it’s seeing the team grow. I’ve been the constant dinosaur in the background with new people like Audrow stepping up to make the podcast better. It’s been really nice to see.
Audrow: Thanks, Peter! Per what have been your favorite moments?
Per: Being able to work with this amazing team, and being able to go out and meet the interviewees and talk to them about what they’re doing their field. I’ve met so many interesting people doing this, and I’m just amazed by all the people I’ve come to know through this work. If anyone out there is listening, doing interviews is an amazing experience: you should try it!
Markus: I’m with Per on this one. For me, it’s all about the people and the people on this team.
I’ve been deeply impressed and inspired by the excitement, energy, dedication and tenaciousness that people have put forward to make this project come alive stay alive for nearly 10 years. Everybody who’s here today has greatly contributed to making this happen and I’m honored to be part of it.
Sabine: What about you Audrow?
Audrow: It’s working with you all. It was an amazing experience going to the same conferences and seeing everyone in person. We were often on calls like this for our weekly meetings, but seeing you in person and connecting, that’s the best part for me.
The final question I have for everyone is: what have you learned from your involvement in the Robots Podcast?
Markus: I think the key thing I’ve learned is that it really matters what people you work with. I’ve long tried to work with people who are better than me, and I think I’ve been extremely fortunate in that respect and I would recommend that to anyone.
Per: Since I’m not a robotics researcher, what I learned from doing interviews is that, when you’re face to face with this person, you’re really trying to be the voice of the listener in that situation. When I listen to an interviewee speaking, I listen in a very focused, structured way – and that’s been tremendously beneficial for me in in general.
Peter: Listening to all the past episodes I notice that the best researchers and most interesting people have a talent for explaining their research in relatively simple terms. I think that’s extremely difficult and it’s a very useful skill to have. In my opinion, if you cannot explain something in relatively simple terms, you probably haven’t really understood it. It’s quite amazing how the best researchers can explain difficult things because they understand them so well.
Sabine: Tying back to what Peter is saying, what I learned is how valuable science communication is. Of all the skills I learned during my PhD, doing the podcast was probably one of the most valuable. Talking to so many researchers about their work opens your mind to all the different areas you might be looking at during your PhD. You have to learn how to ask questions and to master each topic very quickly so that you can talk to them about it. In terms of selling yourself in the real world afterwards, being able to communicate is a key skill, whether you’re in academia or in business.
Hopefully young PhD students will do the same, and will start blogging and podcasting and doing science communication of their own.
Audrow: Alright, thank you everyone. Happy 200th episode!
Part Two: Interview with Rodney Brooks
Sabine: Hi Rod, welcome to the Robots Podcast.
Rodney: Thanks for having me. I’m honored to be on number 200! It’s great that you’ve had so many!
Sabine: Yeah, we’re pretty excited. The podcast has been running for nearly a decade now and it’s been great to have you on this show all these years.
How has robotics changed since that first interview when you were telling us about the subsumption architecture?
Rodney: Well, in the last ten years a bunch of things have changed.
In terms of subsumption robots, there are now at least 14 million of them out there in the world, so that’s pretty good. There weren’t anywhere near that many ten years ago. The iRobot Roomba is really a subsumption-based robot, and it’s been pretty successful in that niche.
I think the other big thing that’s changed is ROS. I love ROS, but it could have easily been another competitor. It’s sort of been a winner-take-all situation. The fact that so many people are using it means that people get to work together and exchange things, and it’s really taken off and grown the research arena. It has also let people see what robots could do, helping them move toward a lot more practical applications than we had ten years ago.
Sabine: I think a lot of our listeners might go back and re-listen to your more recent interview about Rethink Robotics. How is the company doing? Can you tell us about Sawyer and how it is different from Baxter?
Rodney: We had always intended to have more than one robot, but with a common software platform. When you buy most industrial robots, it’s like buying a smart phone. The manufacturer says, “If you want a camera, you can buy that as an extra, and if you want a microphone you can buy that as an extra. Oh, and by the way, it’s possible to write programs so that it can operate as a phone if you write the code.” In other words, most industrial robots come as empty shelves.
We decided to have a software stack with intelligence that works across multiple robots, and we use the analogy that iOS runs on iPads and runs on iPhones.
Baxter was our first robot. We learnt a lot from it, it’s a great robot, and we’re still selling a lot of them. At the same time, we saw an opportunity for a smaller robot, especially in China where it could work side by side in close proximity to people. A lot of people still think that China has so much labor they couldn’t possibly want robots, but if you look at the demographics in China, the absolute number of 19 year olds is dropping by 30% over a 10-year period. Nineteen year olds are where the manufacturing labor force comes from, so China is in desperate straits in terms of labor for manufacturing.
Really good, well-run companies that make electronics for the western world will have a labor turnover rate of 16% per month. You can imagine how hard it is to run a manufacturing company when there’s such high turnover. The nice thing about that problem, though, is that the electronics manufacturers in China have had to break down the manufacturing cost into a lot of simple tasks that a new person can be trained to do in just five minutes. That’s ideal for being able put a robot in there, because it’s not a complex task with lots of judgment involved. It’s a fairly simple task that we can train a Sawyer to do in just a few minutes also, and a Sawyer doesn’t quit after a couple of months.
Sabine: The markets are very different between USA and China. Does that warrant different types of robots?
Rodney: If you go into Chinese factories (and I’ve been going into them for over 20 years), people really are just two or three inches from the next person. A Baxter is way too large to sit at one of those work stations, so we had to build a much smaller robot to go after the Chinese market. We also saw that there was a need to be able to handle five kilograms, which is more than the Baxter payload.
Also, people have hips and can lean far forward, but the robot doesn’t have hips, so we needed a long reach on the arm too. Sawyer’s arm is actually longer than Baxter’s, and it’s longer than Universal Robots’ and the Kuka iiwa as well – this is so that it can do the task that a person with hips can do.
Sabine: It’s really great to see the evolution of these robots. We have a Baxter at the Bristol Robotics lab, and that’s the first thing we show everyone when they come into the lab.
Rodney: Yes, that lab at Bristol is great by the way.
Sabine: As we mentioned, it’s our 200th episode and we’ve been doing science communication for a long time now at the Podcast and at Robohub. I was wondering why you think it’s important for roboticists such as yourself to go out there and spread the word about robotics?
Rodney: I think it’s important so that people can understand what robots can do. We’re seeing in the press a little bit of incorrect information coming from people who are outside robotics and misunderstanding what it can do. The fear of robots taking away all jobs – I think it’s misplaced.
What’s really misplaced is the fear that robots are going to rise up and decide to take over the world. That makes no sense at all to me. But people who are well-known scientists or technologists (but not roboticists) were saying that, and I think it’s doing the whole world a disservice because there’s no chance of anything like that happening anytime soon.
Anyone who works in robotics understands that we can barely get our robots to do most things. They’re not magic machines. I think it’s important for roboticists to talk to the regular press and help them understand that. And it’s important for roboticists to talk to people like you to reach the robot-loving audience and help everyone understand what the significant research problems are that still have to be solved – because there’s plenty of them, and there’s plenty of room for new people coming with new ideas.
I think my role these days is to talk about the plausible things that robots might do in the hope that it will encourage young researchers to come up with brilliant ideas on how to do it.
Sabine: What are these plausible things?
Rodney: I think the thing that’s really going to drive robotics for the next thirty years is eldercare. If you look at demographics in Europe, in North America, in China and even more so in Japan, the ratio of older people to younger people is increasing rapidly. I just saw a statistic yesterday that within a few years a third of all people in Japan are going to be over age 65, and the ratio of people paying into their social security system for every person being supported by social security is going to go from something like 9.5:1 down to less than 2:1 over the next few years.
There are a lot of people who will not be able to pay for the services that they might want, they are living longer, and they – I should include myself – we, we will want our dignity and our independence as we’re living longer, but we’ll need help.
I see the driver assist features of cars turning cars into eldercare robots because it lets people drive safely longer, which lets them have their independence and dignity longer. I also think robots in the home are going to be necessary to help older people.
I was just at iRex, the international robotics exhibition in Japan. There were lot of industrial robots, and a lot of emergency rescue robots following from DARPA Robot Challenge type applications. But it was also interesting to see a whole bunch of booths from various research institutes and universities with robots to help the elderly, to help them with mobility in the house, help them get into bed, out of bed, to help them wash. These robot demonstrations are not going to be deployable in real homes in the next year or three or five, but I think it’s great that researchers in Japan are now turning to this as a place where they can have a real impact.
I think we’re going to need lots of robots to help the elderly so that the rest of society is not completely swamped by this massive relative number of elderly to younger people.
Sabine: I agree that’s a great application. You mentioned the hype and the misinformation. Is that something that’s changed, or has that just always been the case? And what are some of the things that you need to be mindful with your company when you communicate about your robotics today?
Rodney: I think it has changed because of the high profile of some people who have come out and said rather silly things over the last year or two.
But let’s talk about my own obligations at this company. People see our robot and they imagine it can do any task in any factory, and that’s just not true. It can only do a relatively small number of tasks. Our constant thing is to train our sales representatives on what is possible versus what someone would like the robot to do. They have to be able to effectively communicate that, so that people buy the robot with their eyes open to what it really can do for them.
I remember when Baxter was first announced, a prominent blogger said to me, “I’m so disappointed. I thought I was going to be able to have Baxter steer my yacht.” I was thinking, why would you ever think our robot could take the helm and steer a yacht? There are unbelievable expectations of robots, but a general purpose robot – it’s just not going to be able to do something like that for a long, long time.
Sabine: Maybe in another eight years for our 400th episode?! What can we expect to interview you about by that point in time?
Rodney: I think we’re going to start to see some companies really being in the elder care space. I’m a little more sanguine about all the companies in the social robotics space … I don’t know if there’s enough room for all the companies that are saying they’re going to have social robots in the home. I’ve taken issue with the idea of companion robots – I’m not sure that’s where it’s going to be. I think it’s going to be more about physical help robots. Hopefully in eight years we’ll start to see some of that.
I think the robots we’re building with arms that are safe to interact with will let more people do research in that area. One of the reasons I’ve really pushed the team to make Baxter available as a research robot was that I thought it was important to have a platform where people didn’t need to worry about the safety of the arms, so they could start doing work with the robot in whatever field they chose. And I was hoping some would choose elder care. There are quite a few university labs that are using Baxters to do things like help an elderly person get dressed or help someone in a wheelchair, or quadriplegic to feed themselves, etc. If you used a regular old industrial arm, that would be a pretty tricky and unsafe proposition for coming into contact with people.
Even though this company is aimed at putting robots in factories, I think the technology of these safe arms is going to enable the next wave of research and companies. Within the next eight years I think we’re going to see some startups in eldercare begin to appear.
Sabine: Excellent. Thanks, Rod, for joining our 200th episode.
Rodney: Thank you for having me. Congratulations on doing this so well for so long!
Robohub Podcast is a non-profit robotics podcast where we interview experts in robotics, including researchers, entrepreneurs, policy makers, and venture capitalists. Our interviewers are researchers, entrepreneurs, and engineers involved in robotics. Our interviews are technical and, often, get into the details of what we are discussing, but we make an effort to have our interviews understandable to a general audience.