Podcasts Season 2

S02E09: Robotics, Entrepreneurship, and Art, with Damith Herath

In our final episode of season 2, we are grateful to be joined by Damith Herath, Associate Professor of Robotics and Art at the University of Canberra. Damith is a multi-talented roboticist with a long history of working in the art world, and an interest in understanding how to shape human-robot collaboration in real-world environments. During our conversation, Damith talks to us about how his innate drive to experiment with electronics and robotics led him from an entrepreneurial childhood in Sri Lanka to the forefront of robotics and automation research in Australia. 

Listen and subscribe on Apple Podcasts, iHeartRadio, PocketCasts, Spotify and more. Five star ratings and positive reviews on Apple Podcasts help us get the word out, so please, if you enjoy this episode, please share with others and consider leaving us a rating!

Credits:

Guest: Damith Herath (University of Canberra)

Co-hosts: Zena Assaad and Liz Williams

Producers: Zena Assaad, Liz Williams, Robbie Slape, Martin Franklin (East Coast Studio)

Acknowledgements:

A special thanks to the ANU School of Cybernetics for lending us the use of their podcast studio for this recording.

Episode transcript: 

Liz:  Hi everyone, I’m Liz Williams. 

Zena: And I’m Zena Assaad. 

This is Season 2 of the Algorithmic Futures Podcast. 

Liz:

Join us as we talk to technology creators, regulators and dreamers from around the world to learn how complex technologies may shape our environment and societies in the years to come.

Zena:

In this episode, we’re joined by Damith Herath. Damith is an Associate Professor of Robotics and Art at the University of Canberra, and he is also an Honorary Visiting Scholar at the National Facility for Human-Robot Interaction Research at the University of New South Wales. 

Liz:  

During our conversation, Damith talks to us about how his innate drive to experiment with electronics and robotics led him from an entrepreneurial childhood in Sri Lanka to the forefront of robotics and automation research in Australia. 

Zena:  

Damith is an experienced engineer and research academic, but he has also taken the path less travelled. Through serendipitous encounters, he has been a part of multi-disciplinary creative arts projects that combine traditional robotics engineering and artistic explorations of how we, as humans, engage with our creations in different social contexts.  

Liz:

We also talk through his eclectic professional experience and creativity, from his days of conjuring different technology schemes in his youth, to how his fascination with robotics has led him to where he is today. We’re so glad that Damith has been able to join us today.

Liz:       

Thank you so much, Damith for joining us on the Algorithmic Futures Podcast. We’re so excited to be here in person with you in our very first in-person recording of this podcast.

Damith:            

Thank you for having me, Liz and Zena.

Liz:       

Well, one of the things we like to ask people who we invite to our podcast is to tell us a bit of a backstory as to where they started and how they ended up on the path that they’ve ended up on, so I would love to hear a bit about your story, where you started, how you ended up in robotics and art. 

Damith:            

I think that goes back to when I was a kid. My first passion was acting. I loved acting. I was dead serious about actually reading books about method acting when I was about six years old. My dad had a bit of art background. So that’s I think the first foray into art. My dad has been actually an influential figure in I think shaping what I am today. Around ’84 when the microcomputing boom was happening, he managed to acquire a Commodore 16 computer.

In hindsight, I think what he’s alluding to is that I was so much really into art, he probably thought, “Ah, maybe I need to calibrate this a bit, especially art.” Art does not necessarily pay your bills. So, he introduced that. And he also bought a book, one of the Ladybird books. I still have a copy of it, How to Build a Transistor Radio. So, I got hooked with electronics and computer programming. That always been the theme that, I love art — theatre especially, and performing. So, I think being a lecturer, it’s a performance act. Computing, programming, and robotics on one side and art on the other. 

Zena:    

Can I ask, did you ever actually do anything in acting? Were you in any plays? Did you do any commercials? Did you ever pursue anything in that area?

Damith:            

Funny you ask. I did a reasonably good role in one of Sri Lankan soap operas.

Zena:    

We have to find this on YouTube.

Damith:            

So, I’ve been trying to hide that from my son because he’s insistent that I have to see this, and it was not too bad. It was horrible. But before that, I’ve done a lot of acting. I’ve done stage performances in the university. I was nominated once for the best actor role.

Zena:    

Really?

Damith:            

Yes.

Zena:    

What was the award?

Damith:            

It was the Inter-university Drama Competition back home.

Zena:    

That’s so cool.

Damith:            

I was one of the parts in Lysistrata, one of the classic Greek plays.

Zena:    

I feel like we have to link to all of this stuff in the show notes – “To see more of Damith acting in Sri Lankan soap operas, click here.”

Damith:            

And a fun fact, my mom got scared I think after that episode. So, she managed to get me hooked and get me married before things … 

Liz:       

When you went to university, what’d you set out to study? 

Damith:            

It was engineering. As I mature and like I said, my dad must have played a hand in guiding me without explicitly saying, “Look into engineering.” But I really got hooked in with electronics. I was building radios as a kid, and I was repairing neighbours’ radios and electronics. I actually started my first business as an entrepreneur around when I was 12, building Yagi antennas for neighbourhood. So, I was the antenna builder for the local village and the town. 

Zena:    

Wow. How did that migrate into where you are now? You said that you ended up going into university and studying engineering, but how did you… because I know a lot of the work you do now is around robotics and art. So how were you able to migrate your passion back into an engineering pathway?

Damith:            

I did my undergrads in engineering. I wanted to do robotics. That actually has another funny anecdote for that one, because as a seven-year-old kid, I was obviously lazy. Back in the day you had wake up around five in the morning and prep food and all that. So, I always thought, it would be nice to have a robot to do this stuff, because I’ve been watching all the sci-fi. So, one of the early sci-fi is Buck Rogers, I don’t know if any of you are familiar with it.

Liz:       

I’ve heard of it.

Damith:            

Yeah. This is really old sci-fi, and it had a robot called Twiki that brings food and stuff. I wanted to have something like that. I’ve done a lot of stupid things to make that happen over the years since when I was a child. But one of the things or one of the dreams was to know how to build these robots and what it takes. I knew that engineering is the pathway for that. Obviously, a field called robotics didn’t exist then. So, I thought electrical engineering was the way to go.

Zena:    

This might be a bit tangential, but I had a boss once who told me that if you ever have a job that you hate doing, give it to a lazy person to do, and they will find the simplest way to finish that job.

Damith:

Absolutely.

Zena:    

And it just reminded me of what you just said. You said that because you were lazy and didn’t want to get up at 5:00 in the morning to make your food, you were like, “I’m going to find something else to do this job for me.”

Damith:            

The funny thing though, I’m still looking for that solution.

Liz:       

I was going to say, that might be actually the hardest path.

Damith:

That speaks for how difficult robotics is. That keeps me up at night.

Liz:       

So, undergrad in engineering. How’d you end up in Sydney for your PhD?

Damith:            

It’s a series of I think serendipitous interventions, interactions. I never really wanted to go out. The whole thing for me was just, can I build these robots? That’s the only thing. I never really thought of an academic career. That actually didn’t appeal to me. But then obviously, I knew that I need to pursue some higher… Because what I finished in undergrad was in production engineering. So, it’s more a mix of mechanical engineering, electrical engineering and manufacturing technologies.

So, I had these waves of different fields. The next step was to pursue a more focused robotics direction. I was working as a research assistant after finishing my studies in undergraduate, and then it’s really one of the supervisors said, “Damith, I know you are really interested in robotics. One of my colleagues in Sydney is a robotics professor. I might put an email to him and see if he’d like to have me. So that’s how the connections were made, and I ended up coming to UTS to pursue…

So, I didn’t really know anybody in the field. I didn’t know who to talk to or anything. It was just one of my supervisors saying, “That’s the direction you should go.”

Liz:       

And your PhD, what did you end up looking at there? 

Damith:            

If you could imagine the technology we are talking about now in self-driving cars, that actually relies on one key technology called simultaneous localization and mapping. How do you get a robot to understand where a robot is in an unknown environment, and can you actually build a map within that environment? So, it’s kind of like a chicken and egg problem.

My PhD supervisor and some of my colleagues from Sydney Uni actually were the first to show there’s a solution. They accessed a solution for that problem. So, I ended up obviously pursuing that using a certain set of sensors and how we can do that.

This is pre the hype of self-driving cars, and people weren’t really thinking that self-driving cars were a possibility. So, DARPA Grand Challenge came around that time for the first time. Some of the early work around that happened at places like UTS.

Liz:        

Just a post-production note here: Damith refers to simultaneous localization and mapping as SLAM in the rest of our chat.

Liz:       

This is a really nerdy question, but what kind of sensory input are you using to do that mapping or were you at the time for your PhD?

Damith:            

I actually looked at very specific sensors. It’s a narrow baseline studio vision camera. The perception of the day was that studio vision obviously has depth, it’s like our eyes so you can measure the depth. So, it should be able to solve the SLAM problem easily with a studio camera. Because a lot of the work prior to that were using expensive lidar, sonar and so on. So, a lot of the high quality depth camera that we have, I mean, we take it for granted now — it didn’t exist. 

Liz:       

Okay. I know though that you didn’t stay in SLAM. Where’d you end up next?

Damith:            

I had to obviously find alternative pathways to support my research because I would not be ARC competitive, when I first started in art. So, we’ve been very creative about how we go about finding resources. And throughout my career I’ve contributed to building two labs that are really multidisciplinary. So the first one, about a decade ago and now I’m again helping to build a multidisciplinary lab. And I’m really proud that almost all the work that we do within the lab is actually funded through industry partnerships. We are really looking at translational work to help local industry to manifest their products and ideas into real final applications.

It’s funny that Australia has sort of seen that shift – I mean, within the ARC and generally I think the governments, both in the previous and the current government, the trajectories for more translational work. I think that’s good. Obviously, I have reservations that we need to still do fundamental research, but Australia has been lacking in actually doing translational work. A lot of our research ended up being commercialized in the US, for example. So that shift is good. We were doing that long before the government picked that up. It was just out of necessity.

            *** SURVEY CALL OUT ****

Zena: 

Hey there, listeners. We need YOUR input to make our show even better. Your feedback is incredibly valuable, and we’ve put together a quick survey to hear your thoughts on what you love, what you’d like to see more of, and how we can enhance your podcast experience. 

Liz: 

We are asking you for your help because we create this for you, our listeners – and want to make sure what we’re making is valuable to you. 

So, go to our website: algorithmicfutures.org – and you will see a link to the survey there

Zena:  

And now, back to the show. 

**

Liz:       

When you’re considering an industry partner that you would like to work with on one of those types of problems, what are some of the things that you consider in thinking about how to set that partnership up and what that might look like in terms of how it’s structured, and how you’re going to work together?

Damith:            

It’s really important that the partner that’s coming through understands research, that we have a good mutual relationship. That’s number one. I really want to make sure that we understand each other at a really good level, because that sets up for eventual issues and difficulties and problems that might crop up.

And then it’s really, each project is different, and each team and individuals are quite different. And some of the projects are more translation ready, some of them requires a bit more research, so kind of a spectrum. But it’s understanding what that technology readiness level is for each of those ideas. And then understanding what kinds of research need to be brought in. We do a whole host of different things, from setting up little capstone type projects for students, so that we can at a very marginal cost, try different ideas. If they’re much more mature, then we might end up actually hosting a PhD or two to work on that project for say, a three-year term.

A lot of time, one of the understandings is that it’s research, so there’s always potential to fail. We might find other things that we didn’t anticipate. So having that understanding and structuring what kinds of activities happen seems to have worked out well for us. But the key messaging is the communication.

Zena:    

How do you balance the differing expectations between your industry partners and potentially maybe a PhD student? With PhDs, which we’ve all experienced, I often refer to a PhD as a personal manifesto because it really does evolve over the time that you do a PhD, and it’s a very individual and siloed project. If you have PhD students working with industry, doing an industry project, how do you balance the difference between the expectations of a PhD student who wants to be able to mould this particular research in their own direction and claim their own identity to it, versus an industry partner who has a set of outputs that they’re expecting?

Damith:            

That’s a really good question and something that I’m still grappling with to some extent. We’ve found measures where we can mitigate some of that. One of the first students actually that started in the lab really didn’t understand the value addition the industry partner bringing in. At some point, it actually became contentious between the industry partner and the research student. And one point the student asked, “Why am I doing free work for this industry partner?”

So as a supervisor I can see both points of view and I can see where the student can’t really see because lack of experience. They come with a certain expectation, like you said, it’s an individual enterprise. And in that, why would I commit to a project that somebody else benefits? It’s always an interesting question.          

One other way we’re mitigating it now is that we have very long conversations with potential students who want to come on board with the industry partners to go through a number of iterations, discussing what each party expects, and making sure that we meet in the middle. I’ve been lucky because all the industry partners I’ve got so far have been very open about the intellectual property, the way the IP is handled. They’re really open to the idea that, the students should be publishing, they can publish. So, there’s a lot of flexibility in that. Been reasonably lucky until now.

Zena:    

No, I appreciate your honesty because I think that’s the reality. There is no solution. There’s not usually one answer. And what works for one dynamic may might not work for… What works between one research student and one industry partner might not work for another dynamic.

Liz:       

I’m remembering back to the first time that we met, and I came and toured your lab. You had one of your projects with Stelarc set up that one of your students was working on. You had another project where one of your PhD students was working on this robotic arm for recycling sorting or clothing sorting. And then you had something else a student was working on that was going to be for a hospital setting. It was really interesting to see the breadth of the projects that you were doing.

And so, I was curious, is there a common theme in terms of the kinds of projects you choose to work on, or you choose to bring students in on? And how do you go about deciding on what to focus on?

Damith:

The short answer is being lazy, because obviously all of the things that I do, centrally the theme is in how do you get robots to actually do anything? That actually requires a whole breadth of things like that needs to be happening. So, the examples you mentioned, they start for the Stelarc’s project, this is an art project that I worked with Stelarc, the artist I mentioned earlier. That’s about understanding how humans relate to machines and interact with them.

And David’s PhD project, which he’s going to commercialize shortly, is actually developing a gripper, basically a hand for a robot to manipulate clothes in this case. That’s another aspect of it. The hospital work is again understanding some of the cognitive architectures that’s useful to intervene in human machine interaction domain. And again, in that is a plethora of different ideas and there are topics from computer vision on one side, to natural language processing and again interaction and body conversational systems and so on.

I think the common theme is really trying to understand how to go to build these useful machines that can do things for us, but in a more meaningful way. Because right now the way I treat robots, they’re really tools. However much you want to anthropomorphize machines, robots, they’re still tools. There should be a level at which robots become a bit more often collaborative. I know Liz, you work also looking at these aspects of collaborating machines.

So, to build that I think you need to build all these different technologies, understanding the psychology of it, the design side of things, and of course, the mechanical and mechatronic aspects. So, I think that’s my goal is that childhood dream, to build a robot one that can look after all that stuff that I don’t want to do.

Liz:       

Do you think that your work in robotics makes it easier for you to see robots as tools as opposed to somebody who is maybe looking at robots from the outside and considering how they might become a part of our society, culture, or world in the future?

Damith:            

I still find robots to be tools. And some of the earlier research we actually did around trying to understand the perception of machines indicated that, generally people see them as tools, or at least they only would accept a robot if they have a utilitarian purpose. So that romantic notion of a machine being a lover or a partner or whatever, I think it’s a bit too further down. But at the same time, humans are easy to fool. So, you could in the short term, fool people into these kinds of relationships. But personally, I think they’re useful tools.

So, the question is how we can adapt current existing technology to make them much more user-friendly. The analogy I use is the smartphone. Now a smartphone, a six month old kid could figure out the interface and go to YouTube and play a video. And that’s a lot of research that’s gone into making the interface simple and useful. We don’t know what the robotics equivalent of that. So, I think that’s what I’m actually doing to find these utilitarian interfaces that help humans to do their work better, efficiently, and in a more meaningful way.

Zena:    

I just have a quick follow up question to that and it’s on the overreliance of robotics. [00:43:00] The example I want to use is my husband and I during a Black Friday sales last year bought a robo vacuum. When we first bought it, we said that we were going to use the robo vacuum throughout the week, but still at least once week or once a fortnight we’d have to get the proper vacuum and mop out to get the corners and all of that. Literally hasn’t happened. Have not touched the actual vacuum cleaner since we got this robo vac, because we’re like, “Oh, it’s doing a great job.” It’s not. I bet if I go into the corners of my house, I’m going to find missed spots.

So, it’s a long-winded question of being like, is there some kind of barrier or maybe consideration that we have to think about when we’re building these systems when it comes to the overreliance on these systems? I’ve used a pretty vanilla example of a robo vacuum in my house, but if we’re thinking about other examples, and I can’t think of a really good one now, but is there some level of consideration that you put towards overreliance on these systems when you look at projects that are looking at building them for different purposes?

Damith:            

I would differ though. I actually for the longest time, deferred buying a vacuum cleaning robot, but I did, like you-

Zena:    

And you can’t look back now. Yeah?

Damith:

Let me tell you. I actually bought a vacuum cleaner recently off one of those sales as well. It was a real bargain, it was 50% off. So, I thought the economics work. And somebody who has worked in SLAM, I find the mapping is substantively improved over the years. I was impressed. I don’t know what your brand is, but the one I got, it does a really beautiful map of the house.

Zena:    

They just miss the corners because it’s circular in shape and your corners are perpendicular. They physically cannot get to a corner.

Damith:            

I agree on that. But from point of view of the technology-

Zena:    

It’s fantastic.

Damith:            

it’s fantastic. To do a map like that, when I was doing my PhD, that was impossible. It was really a state of the art if you could have done that back then. So, the SLAM, the area and size of what I’ve done is about 10 by 10, so 100 square meters, and as good as what these robots can do. And also, they’ve improved the utility of it, the suction powers up better. It does a really decent job. So, I haven’t really actually used the normal vacuum cleaner for a long time because it does a decent job. And I agree that they’re probably still cutting corners.

So, it’s a good question. Actually, I was listening to Rodney Brooks, one of the inventors of the original vacuum cleaner, the Roomba vacuum, one of the two vacuum cleaners, Roomba vacuum. He said after the Roomba or the vacuum cleaner, there’s absolutely no other robotic artifact or product that has come out of the market that’s actually survived. We’ve seen Jibo and all sorts of other social robots, but none of them really survived the retail market. And so that’s one of the questions I have actually. It was a really good question. I’m trying to think of the things we are doing, what makes sense to do, and not to over-promise.

It’s funny because we made some critique about one of the social robots that came out and said, “Based on our research, if they’ll not succeed commercially for X, Y, Z reasons.” But then we got a really brutal rebuttal from the reviewer saying, “These are really highly respected researchers doing this work. How do you criticize that?” And lo and behold, the product actually failed, and people complained that they’re not delivering on the promises or expectations.

We did a project called Adopt-a-Robot a long time ago. The idea is if you have a child, a toddler, and as they grow, you create a bond with them because they do very interesting things over time and they improve communication and so on. So, we said the same thing that if we had a robot that’s really random to start with, and then if we improved its communication abilities and things like that, people would bond and have a higher attachment to the machine, which didn’t happen. So, we were improving this robot over three-month intervals, and it just stayed the same.

And the conclusion that we came to is that if there’s no strong utility for [00:48:00] these machines, people lose interest over time. Even now I’m really looking for that next robot that actually would have that utilitarian outcome for people.

Liz:       

It got me thinking about the fact that I will never have a robot vacuum at least as long as I have little kids because there’s no way they would ever be able to make it through my house.

Zena:    

You have to pick everything up the floor.

Damith:            

Hang on. The newer ones actually detect poop and cables, so you don’t have to take anything off.

Zena:    

Poop and cables?

Damith:

Yes.

Liz:       

That’s amazing. Essentials.

Damith:            

Yeah. They’re all apparently tangled free.

Liz:       

I wanted to get a sense of what it’s like to have one collaboration with an artist. I’m wondering if you could pick a project or a collaboration in the artistic space that you’ve worked on and talk about how these projects evolve over time. What is your role? What does the creative process look like?

Damith:

Starting in the beginning, it’s a huge learning curve, because I am trained as an engineer. When it comes to engineering side of things, I have a very logical thought process, how to build these systems. And when it doesn’t align with that, I go, “What’s going on here?” So, working with Stelarc, in a way it was useful because he has done these kinds of work for 30 years, and had worked with a number of really prominent engineers. So, he knows how to manoeuvre that, and that helped me to really learn the ropes.

One of the things that I learned over the years is the language. We sometimes end up talking about the same thing but use different language. And sometimes the goals really do not align. One of the examples I regularly use that Stelarc says, and Stelarc says that his career was built on failed projects or failure, and one of his edicts is that “Damith, make mistakes.” And that kind of antithesis for engineering, you obviously don’t want to make mistakes, especially if you’re building a bridge. So, understanding and appreciating that the creative process involves going on tangents, making mistakes.

A lot of the early work actually was a result of just me making silly engineering mistakes and Stelarc going, “Oh, that’s interesting. How did you do that?” And I said, “I don’t know, I mean, it just didn’t work. Let me look into the code and see what’s going on.” And that’s how you build that rapport and understanding. And as you mature, then you understand the common language that we can use together to express ideas and intentions. So, it becomes a bit easier.

I had actually a conversation with Stelarc, it will come out as a paper shortly, where we discussed the journey over the years. And one of the conclusions that we came at the end of that conversation is that maybe we will never be able to reconcile these two fields working together. That’s fine. The artists will have certain expectations and interests and drivers, and the engineers will have certain different ones. But that’s a good thing. We don’t need to work towards a common goal. You can have different goals.

Just an example, we might want to write some algorithms and talk about them in our publications. For Stelarc, it would be a public performance, and that works in different level, and what do we do is a different level. As long as you can find that commonalities or different goals to meet, I think that’s the way to go. The important thing is to be open-minded. I think if anything, I could say, you need to be really open about the possibilities of the other, and respectful of each other’s intent and where they’re coming from.

A lot of the projects I think fail at some stages because these expectations don’t meet and people realize, “This is not actually going to help me advance my work or career.” And that’s unfortunate.

Zena:    

You and your collaborator, David St-Onge have just released a textbook, Foundations of Robotics, a Multi-disciplinary Approach with Python and ROS, which we’ll link in the show notes for our listeners.

So last we checked, this book has 120,000 downloads, which is incredible. And you’re also leading the development of a specialist major in robotics and AI at the University of Canberra. Can you tell us a bit about your approach to educating students with an interest in robotics, and what you think of as foundational knowledge for the students that we’re educating now?

Damith:

It’s been a dream, actually. That’s one of the other childhood dreams. I want to actually write a book, not this one, but I want to write a science fiction book that’s-

Zena:    

Oh, that’s such a cool dream.

Damith:

In fact, because you picked… Hopefully I have a bit of time to talk to this story. When I was a kid, I once entered into this writing competition. I wrote a novel about… I don’t know how I got this idea because it’s long before the internet and everything else. But I talked about a robbery happening at a bank where a person using a computer actually managed to get the money out. Rather than actually, somebody breaking into the bank, somebody can go online. I didn’t use those words. In hindsight those are the words that could’ve have been used.

But the narrative is that there was a thief who could access a computer in this bank at a time when banks actually didn’t have computers. So, I’m talking about eighties. Where I come from, only I think probably the head bank would have had a computer.

So, writing a book was always dream. And that dream came true because when we wrote the Robots and Art, edited the Robots and Art book, that took about five years.

But I’m really proud of that one because we’ve managed to bring in a number of different people from specifically two different disciplines together for the first time. And some of the seminal work in robotic art was there. And the artists themselves actually talking about the work. So that was my first book that I’m really, really proud of.

This one is really out of necessity, because when I started looking at developing a new honours program for our university, I want to have something that empathize with what I was thinking about working in multidisciplinary spaces. And I’m finding increasingly robotics is multidisciplinary. It requires much more broader thinking. This is I think the problem we had with AI because the multidisciplinary context didn’t really, I think come through well. There’s obviously been media arts and things like that, but it wasn’t really a theme when a lot of the work happened.

We want to see a paradigm shift in how we think about robotics because robotics is still really male dominated, very focused on that hardcore engineering. And if you go to any research conference in robotics, you’d see that kind of quality to it. People don’t really understand anything left or right of that, very narrow focus. So, we already deliver up a course that actually addresses that bit. So that’s one thing. The other one is when we are looking at what the content should be, we found that there’s completely different approaches, depending on what origins you had.

If you’re coming from computer science, computer engineering, there’s a specific tone to the course. If you come from mechanical engineering, there’s a very strong mechanical component and the machine learning aspects not really prominent. So, we want to define a body of knowledge that we think is useful for modern robotics. And we obviously interact with a lot of industry partners, and we were listening to know what that book should look like.

So, the book is really a result of those conversations and trying to come up with a body of knowledge for robotics, which we think is comprehensive, multidisciplinary, and expanding on some of the emerging topics like ethics, design, psychology included in it. That’s how the book came about. But what I’m really proud is that we managed to have an industry partner to sponsor the book to make it open access. Kinova Robotics kindly sponsored the book, one of the prominent robotics companies in Canada. So that means the book is accessible to students. Textbooks are pretty expensive as well.

Not only that. Each chapter actually contains insight from an industry veteran. So, one of the other things a lot of students complain is like, “You’re teaching all this stuff. Is this relevant in the industry?” So, each chapter actually connects with industry expert who connects their own journeys to what they’re doing and how that chapter resonates with them. So, each chapter has these industry partners talking about their own engineering pathways and how that connects the chapter. So, it’s really a comprehensive introductory book also relating to how industry practices align with the actual study.

Zena:    

I think what you’re talking about with the different engineering pathways really resonates with me, because I wouldn’t consider myself a tinkering engineer. I don’t particularly gravitate towards physically building things. The engineering I’ve been involved in has been predominantly systems engineering and safety management systems of emerging technology. And often, when I tell someone I’m an engineer, they automatically think that I build things. And then when I clarify and say I’m an aerospace engineer, they think I work in aircraft manufacturing or aircraft design. And I don’t work in either of those things.

So, for me, it really resonates with me to see something that is representative of the broader spectrum that is engineering, which extends really well beyond just physically tinkering with things. Not that there’s anything wrong with that. They all have their value, but I think it’s important to have that representation.

Damith:

Absolutely. Having said that, the book has the most important theoretical aspects, the hardcore engineering in there. But what we’ve done is to integrate that with the other softer components that actually makes robots useful things for today.

Zena:    

That’s awesome.

Liz:       

That is awesome. What about the program that you’ve been putting together, the major… I know you’re building a new lab and all of that kind of thing. What’s your aim with that as you’re building it up? What would you like to see?

Damith:

We want to have well-rounded robotics engineers coming out of that program that are really proud to acknowledge that they understand a broader narrative of robotics that can easily converse with someone like Zena who can understand how engineering is not just about tinkering, it’s about a number of other things as well. And they can then very easily venture into any of these different areas.

We’ve been talking to a number of, again, industry partners and other researchers about this in the course. Just to give an example, we’ve been talking to Amazon Robotics, what they think about our course, the structure and all. That’s actually wonderful because a lot of students coming out of well-established institutions don’t have that breadth of knowledge, and they’re very single minded about their approaches. And it basically makes it difficult for them to adapt to some of the more demanding situations in a commercial setting like Amazon.

So that’s what I’m proud about that. Obviously still in development, but hopefully the students who come out of it are much more self-aware and understand and can be empathetic about the needs of the end users.

Liz:       

Is there anything we haven’t touched on that you want to share with our listeners?

Damith:            

I’ll tell one last story maybe from my-

Zena:    

I’ve been loving these stories. It’s so good.

Damith:            

One last story about one of the projects I did as a child. I talked about the Twiki robot from Buck Rogers. This is a humanoid robot, slightly taller than the Pepper robot that’s popular these days. It’s a humanoid form. I wanted to actually build this thing when I was about 13 or 14 years old. Again, you just need to remind that there’s no internet, there’s no YouTube to figure these things out.

So, a mate of mine and we said, “Well, let’s build this.” He’s a good carver, he could carve wood. He did a lot of the body parts using wood. So, he found a soft wood. And then for the arms and stuff like that, we had Rigi foam, little things carved out again from foam. So, we put the whole thing together, and then we want to make it move back and forth because building an articulated robot is extremely hard, which we figured out pretty early on. It was like, it would be nice to just move it back and forth and that’ll be cool. Then we need to figure out talk and all that sort of stuff.

Eventually found a motor from a car that put it underneath the robot and so you can move back and forth. And then we want to have it to be able to talk. And so, what we did, we built a little amplifier and put the speakers and the electronics inside the robot’s head and two LEDs, so when it speaks, the lights light up.

It was for exhibition. So, we took the robot to this exhibition and my mate would hide inside a box with a mic. And whoever wants to talk to the robot, they had to pay 25 cents and ask one question. So, we ran this scam. People actually believed that the robot could talk and obviously all these kids, same age, until our teacher found out that we actually been scamming all our mates, and we had to shut down the whole experiment–not before we made some money and go to the tuck shop and get some sweets.

But the important thing is I realize that actually, the same thing the research are still doing, and it has a name. It’s called Wizard of Oz studies. So, any robot that can’t really perform a particular action, but you want to test that hypothesis, you’ll get a wizard, like in the novel it’s Wizard of Oz. So, behind the curtains and actually control the robot to action that and test the hypothesis. I was probably the first one to do a Wizard of Oz study with robots in 1984.

Zena:    

You were ahead of the curve.

Liz:       

That’s awesome.

Zena:    

That’s an amazing story. I’m also picturing how small or big this box was that your friend was hiding in.

Damith:            

We were pretty small then, so I don’t know, one meter by meter kind of box. But actually, it was interesting to see people react to that even at that age to believe these things. And that has been a theme actually with a lot of the robotic artwork and something that I’m really against now, doing that kind of experiments because A), you’re saying that you don’t really have the technology, but you want to test these kinds of hypotheses that are probably not useful.

A lot of the research we do in HRI, and this has been a criticism rejected against HRI, human interaction research, is that a lot of the work that’s been done is frivolous and not useful, which is something that I empathize with as well. We try to veer away from that as much as possible, to study robots in real life situations with real technologies available to us today, and understand what that means, implementing them right now.

Liz:       

In terms of future design then, how do you think through what kind of experiments or design processes might actually inform that future?

Damith:            

Well, the current thinking is that we want to do in the wild research, in the wild experiments. I’ve been actually fortunate to have access to places like Questacon and other public situations to run some of these experiments. So, you get really real-life experiences.

If I may share some recent experience, we were running this experiment at the Questacon. It’s the Baxter, the robot. This is a two-armed, gigantic robot, slightly taller than a human. It doesn’t really have a screen for a face. Two big arms. And it was playing tic-tac-toe with a participant. And it was playing for real. We developed the algorithm so it can actually detect the moves and play for real. So, everything was real, it’s nothing Wizard of Oz.

But in one condition what we did is to manipulate it a little–this is where the Wizard of Oz part comes in–to cheat. So occasionally the robot will cheat. So, what we observed in this very public setting, people are allowed to come and play with the robot and it’s a very public open space. A lot of times the families will come and play. When the robot is just playing as a robot, nobody really pays attention to the robot, they just play the game and then move on. It’s just like you’re playing with a computer. They don’t really think about the mechanism, they don’t think about the embodiment.

But whenever the robot cheats, immediately they start looking at each other and started looking at the robot and calling it names and suddenly it becomes he or she or whatever they imagine the robot to be. And so like, “Oh, you’re cheating,” and smiling and looking at each other, and these interactions start to happen.

The point is that if you just constructed this in a lab environment, as a lot of the HRI work does and try to control the variables, so you’re just manipulating one variable, making sure that’s the only one being tested. You might have not those interactions because it’s that kind of human connection that triggers that sort of reactions.

That’s what we want to do, really situate these problems in real world scenarios, try to understand what these kinds of interfaces should look like, what informs the agency of a machine to humans, and how do you use that in actually building real life systems? I find those interactions and situations really interesting. We want to really push into trying robots out in the wild and testing them out and see what people think about them.

Zena:    

I’m super interested in the uncooperative side of this. I think a lot of the times when we test robots in the wild or even just systems in general in the wild, they’re usually cooperative, they’re usually shed in a positive light, they’re supposed to be contributing to something. So, the fact that it was cheating and it’s playing this uncooperative role is very intriguing to me.

Damith:

Robots have always been uncooperative.

Liz:       

It’s quite a natural state. Right?

Damith:            

Yeah, absolutely. People ask because you’ve got all these robots, “They must be helping out and things.” Absolutely not. And the answer to the question, would robots take over your jobs? Absolutely not. For each robot we need about five or six engineers and other people to maintain it and just to keep it running.

Zena:    

But isn’t it interesting that the narrative is so… And I think it’s part of wanting to promote implementation, because if you come in with a narrative of these are uncooperative and you need five people to manage them, you’re less likely to get uptake. So, it’s really interesting to me that the narrative and the reality are, well not quite different, but maybe just not as aligned as they should be.

Damith:            

I think people are coming to realizations that robots could be seen as productive partners. And I think that’s a useful narrative as researchers and as engineers, to explore further. Because the early robots are really tools, and obviously they automated certain areas of our work and got rid of people. And that’s our, I think, experience so far. But the ones that have been developed now, they really require human intervention and interaction. So, we talk about human in the loop kinds of robots. And they, by design will require humans to intervene and collaborate. So that’s a different paradigm from the one that we’ve been exposed to. That requires a lot more work to be done. I think people are starting to realize that. I think the notation would be to explore robots as collaborative partners than replacements or tools, or whatnot.

Liz:       

Thank you so much for listening to our final episode of season 2. We have had an amazing line-up this year, so if you missed out on some of our guests, head over to our website, algorithmicfutures.org, or – even better — subscribe to our podcast in your favourite podcast player. 

And finally, if you are enjoying this content, please leave us some feedback and a five-star rating on Apple podcasts. We love to hear from you!

Leave a Reply

Your email address will not be published. Required fields are marked *