Bill Smart, Professor of Mechanical Engineering and Robotics at Oregon State University, helped start competitions as part of ICRA. In this episode, Bill dives into the high-level decisions involved with creating a meaningful competition. The conversation explores how competitions are there to showcase research, potential ideas for future competitions, the exciting phase of robotics we are currently in, and the intersection of robotics, ethics, and law.
Dr. Smart does research in the areas of robotics and machine learning. In robotics, Smart is particularly interested in improving the interactions between people and robots; enabling robots to be self-sufficient for weeks and months at a time; and determining how they can be used as personal assistants for people with severe motor disabilities. In machine learning, Smart is interested in developing strategies for teaching robots to act effectively (or even optimally), based on long-term interactions with the world and given intermittent and at times incorrect feedback on their performance.
kegan: [00:00:00] Hi, welcome to the robo hub podcast. Would you mind introducing yourself for us please?
Dr. Bill Smart: Sure. Uh, I’m Bill Smart. I’m a professor in robotics at Oregon State University and for the past year, I’ve also been part-time at Amazon. There as an Amazon scholar.
kegan: Awesome. So have you always seen yourself working in robotics and CS and STEM or, or sort of, what’s your story getting to where you are now?
Dr. Bill Smart: I’ve always sort of been a science guy, a math guy. I was not good at math. Uh, mathematics in high school kinda fell into computer science from there. And then from computer science into robotics. The origin story is a little weird. I first thought about robotics cause I saw an article about Rod Brooks and his group and uh, I think it was Smithsonian magazine and it just looked really cool and that sort of led me to robotics.
kegan: I feel like a lot of us in robotics get into it just because it, it [00:01:00] does look cool. It’s visually cool. and it’s something you can build and actually have a an actual thing from the computer science, sort of, side which often is just, you know, software. So, tell us, how have you been involved in competitions because we hear that you sort of helped start the, ICRA competitions.
Dr. Bill Smart: Yeah. Um, so I think it was 2008 in Pasadena. Um, we had the first ICRA robot competition or something, Paul Rybski who was at CMU then, and I, led for a couple of years, we got a lot of support from ICRA. Um, I think Gaurav Sukhatme was the general chair that year, or he, he gave us a lot of support to get it going. And really, it was just an effort to try and get robots at a robotics conference, ICRA a great conference and there’s a lot of great work there, but the only robots, historically, were in the, the vendor booths. And it, it felt like we should have some. They kind of [00:02:00] tied the robots to the, the research, being presented.
kegan: Yeah, so what were the original competitions, like what were they, what were you trying to push? What was it doing? Oh,
Dr. Bill Smart: that’s, that’s great. So there’s a real tension with competitions where you want to reflect what’s going on in research and not make it a off special thing that diverts people from research because then people don’t really get interested in it. Um, the first year we had. Well, I, maybe I should’ve done my homework. First year. We had it was like a planetary exploration thing where we built a little Mars scape from gravel and stuff in the, in the convention center. I think there was a modular robots event. Mark was involved in doing a modular robots.
And something else, which I can’t, I really get that’s fine. But what we did is we put out a call for people who were interested in running a competition. So we facilitated it and helped with the infrastructure, but we had people [00:03:00] propose competitions and run them independently. [Hmm that’s cool]. So I’m trying to reflect what was so hot in the research field.
kegan: Yeah. I feel like competitions are kind of this in-between ground of industry and academia too. And you sort of mentioned that you had you’ve worked briefly at Amazon, right. Uh, or you spent some time there. How do you sort of see that relationship between industry and academia or maybe tied in with competitions as well?
Dr. Bill Smart: Well, I think in the history of the ICRA competitions, it’s been mostly academics. Who’ve led them, but a couple of times I know Amazon led the picking challenge, which was a competition for a few years and Fetch robotics led the I forget what they call it. The Fetch manipulation challenge for a couple of years, I think it was a way to, you know, for companies to engage with the academic community in a relatively structured way.
You know, maybe they have a problem they want to solve. Um, or a problem they want to inspire people to think about. [00:04:00] Um, I think it’s hard to use the stuff that comes out of competitions directly in industry because of IP problems and because just, you know, things have to be a lot more hardened in industry, I think, but I think it’s a good way to get that first engagement, get people interested in that, you know, in, in a part of the space maybe.
kegan: Definitely. Cause it also is something, you know, sometimes visually entertaining with like robotic soccer or something like that, if that’s sort of the competition and I can see that being really engaging.
Dr. Bill Smart: yeah. And I, I think the, they grow competition, you bring up Robocup, right. And I think Robocup has been tremendously successful, but it sort of focused on the competition and the science grew out of the competition.
And I think what we were trying to do at ICRA is kind of maybe the other way around where we tried to get the science to, to use the competition to demonstrate the science as instead of having the competition drive the science. If that makes sense.
kegan: Yeah. As a way to sort of show off the, [00:05:00] the research that’s been going on here.
Dr. Bill Smart: Yeah. And I get all the tangible way of actually seeing the robots in action and maybe, you know, force us as the researchers to be a little more honest about the assumptions we’re making. When we formulate things in our paper.
kegan: That’s another good point. It kind of gives you also a deadline.
It kind of gives you like something to push towards something to work towards. Whereas when you’re working on research, it can kind of be a little nebulous in how you set your own goals. Yeah.
Dr. Bill Smart: I think, I think it’s a good thing and a bad thing. I think the deadline is a good thing, but. Uh, the flip side of that is you have a deadline, which is right next to the competition and right next to the conference rather.
And when you go to ICRA, you kind of want to go and see the talks because the talks are good. Um, but if you’re in the competition, there’s sort of a tension there of like, unless you’ve met the deadline already, no one ever does, like, you’re kind of hacking away at midnight and, you know, do you do the competition or do you do the do the conference.
I think we’ve, that’s been a challenge over the years cause you [00:06:00] want it to reflect the research, but if you’re reflecting this research, you kind of want to go and present and listen to research talks at the conference. And so I think that’s something we’ve been working through over the years to try not trying to find that right balance.
kegan: Yeah. And so have you seen the competition’s kind of evolve over the years and change?
Dr. Bill Smart: Yeah, I mean, certainly the, the material in them, the subject matter is changing. Um, I think it’s still. They’re still not as integrated with the main conference as they might be. And I don’t really know how to do that.
Um, but I thought I’m really pleased to see that it’s still going, like this is 2022 and we have competitions this year. Um, and so it’s really cool to see what Paul and I pitched out there, you know? It feels like a thousand years ago. Um, it’s it’s still going, so there must be some utility in I
Have you seen any sort of real-world benefits that come out of it or any, or do you have [00:07:00] any.
Dr. Bill Smart: Um, well, I mean, I think all this, I think the engagement with companies that are here, if you’re a company at ICRA you can come and see someone working with robots and actually see tangibly, what they’re, they’re up to, rather than just listening to them present.
Maybe, maybe there’s something that’s coming out of that it’s so hard to quantify that maybe I think it’s, it’s got people interested in different parts of the space. And then maybe that has led to some results, but I think the competitions themselves, I, I’m not so sure the, how, how to direct, then you can draw a direct line between the competition into something in the, in the world.
kegan: Yeah. It could be hard to sort of make that direct
Dr. Bill Smart: at the same, at the same time you can think of, you know, asking the same question of papers. Like if you see a paper, right. Huh? Yeah, a lot of the papers you see at ICRA, make it into product and make it into systems. But the, you know, the line, isn’t all always obvious.
kegan: I one thing is I know that some of your work [00:08:00] you’ve done and correct me if I’m wrong, but you’ve worked on having longer term research with, with like an experiment. Right. Um, yeah. So competition feels very short-term. Do you think that could ever be a long-term competition.
Dr. Bill Smart: That’s a great question. Um, in some sense, Robocup does this, you know, because you, you release your code and people build on your code, I guess. Yeah. Year on year. Um, but could you have the same thing where you come back every year? Um, that’s a really interesting idea. I’ve never really thought of that.
I think logistically, it’s hard because you get quite a lot of churn within the student body. If you’ve done it this year, and you’re writing your dissertation next year. You’re, you’re going to be less inclined to do it. Um, that’s a really interesting question though. Cause then that would be a great way to demonstrate progress year on year, which is a thing, one of the things you want to try and do originally with competitions.
kegan: Yeah. One thing. It could be hard to get people or maybe [00:09:00] not, but get people involved if you’re already doing it like a multi-year thing, because normally I would think. Everyone like new people every year to be joining right. The competition.
Dr. Bill Smart: Yeah. Again, I, I, again, I think you want both things you know, the new people, you know, it’s concentrating for new students and it gets more people excited.
Um, but you still need that, that year to year knowledge transfer if you’re going to be building on things. Um, I think the challenge is an academic is funding competitions year on year, ’cause, you know, you have a team of five students, you have to, you know, wherever ICRA is, and ICRA goes around the world, you have to find funding to get them and the robots to ICRA reliably to, to sustain that.
Um, it’s a really interesting idea though. Yeah. Uh, if anyone’s listening and wants to pitch it, I would, I would love to support that.
kegan: Yeah. So there we go. If anyone’s listening and wants to do it so I see competition is often sort of being an [00:10:00] interdisciplinary approach with multiple people working on a team to do something. And that’s also just true of robotics generally. Yes. Um, one thing I was intrigued when I looked into your background was that you have worked with robotics and policy and sort of how the two overlap. Maybe. Could you talk a little about that and, what you’re interest there is.
Dr. Bill Smart: Yeah. Um, so we’re, we’re starting to see robots coming into the real world.
Um, you know, you can, you can actually buy robots and have them in your home. Now. Um, industry’s using robots extensively on one of the things that is causing concern in an industry, I think is litigation over robots, how the law interacts with robots, how they’re regulated what will happen when things go wrong, if there’s an accident.
Um, and there’s just a lot new material. In that space, right? We haven’t litigated a lot of robotics cases. We don’t have many legal scholars, legal practitioners, policymakers who know about [00:11:00] robots. And we don’t have a lot of roboticists who know about the law and know about policy. Um, so about 10 years ago with a co-author Neil Richards, who’s a law professor at WashU in St. Louis we wrote a paper, trying to draw some outlines of how the law should start to think about these new technologies. And so it draws heavily on a lot of scholarship in the legal community already around the, the, the internet and other technologies, a lot of stuff influenced by Ryan Calo from University of Washington.
Um, and we, we managed to actually get that into a conference called WeRobot which is this September in Seattle (if anyone wants to go) which is a great conference that brings together legal scholars, policy scholars, practitioners, and technologists, and talks about real practical problems and real practical solutions at the intersection of all those things.
And it just struck me as a) there aren’t enough technologists. Uh, there aren’t enough [00:12:00] roboticists engaging with legal community and b) I’d really like to learn more about the legal community. And so it’s just been a really interesting way for me to learn stuff that I didn’t learn about and soft to expand what I’m thinking about in my day to day.
kegan: That’s really cool! I feel that often we, it’s easy to be like, oh yeah, the people working in law don’t know anything about robotics and people in robotics kind of assume that, oh, the politicians, they don’t sort of know anything about what we actually work on. Um, but I, I don’t really know what they’re working on or how they, I don’t really understand the whole process. And so that is cool to learn both sides. Uh, both directions.
Dr. Bill Smart: That first year that we robot conference, I was in the back row on Wikipedia, looking up all these really simple legal terms because I, I really didn’t understand it, I think. And over the years, I’ve, I’ve got, I’ve got some understanding of it now, but I think the best thing about going to conferences like that, it’s not that I can learn the law it’s [00:13:00] that I can learn enough of the grown-up words to talk to people who know the law, and we can have a real conversation, a meaningful conversation, and maybe I can help get them grounded in. You know, what robots are and what they’re not, and what science fiction and what’s practical. And there’s a growing community of roboticists who you would recognize from ICRA, engaging with communities like this. And I think I would encourage anyone listening to get involved in that because it’s, it’s really interesting and I think, the only way we’re going to get good regulation and good policy and good law is by having both of these sides really get into it and have these deeper conversations. Yeah.
kegan: Do you think a competition could ever exist there to help them really get into it? Um,
Dr. Bill Smart: yeah, potentially. I think the, you know, this year we have a competition that sort of focused on robot.
Hmm, where, you know, part of it is you propose a design and then there’s, there’s going to be a little [00:14:00] hackathon at ICRA to try and implement some practical ethics. Um, and mostly to understand, not to come, come with, come up with a complete system, but to understand how, how hard that is taking the way that ethicists and lawyers talk about things and grinding it out and working code.
Um, I think one of the problems of engaging the legal community to pick, to pick that community with competitions is a problem we’ve had with competitions from the start in that you’re not really rewarded for winning a competition, like as a faculty member or as a grad student, you’re rewarded for getting papers or maybe getting funding or your new results.
And so I think one of the things. That’s been hard in the legal work is trying to like line up the incentive structures. So the papers that I’ve published in the legal community are technically [00:15:00] unpublished from my, my promotion point of view, because there are evaluated by undergraduates in law journals.
They’re not peer reviewed in the way that we view. Yeah. And so the incentives for me are like off from the incentives from the legal community. And I think that carries across doubly in these comp competition competitions.
kegan: Interesting. But it’s obviously something beneficial to do. So it should somehow, you know, you would hope.
Dr. Bill Smart: Yeah. I don’t think we find ways to do it. Um, you know, you, you write a paper maybe with more of a slant towards the technical community and you publish it. Uh, maybe not in, in ICRA, you know, other, other conferences, like Advanced robotics and social impact, something like that for AIES. Um, and then maybe you write a slightly different paper, but it’s more slanted towards the legal community and you publish that in a law review or something like that.
But it’s something I think it’s easier for a roboticist is easier [00:16:00] later in your career. Maybe when you’ve, you know, you’ve, you’ve got a background of technical publications that your tenure committee is going to be happy with. It’s easier to do. Yeah. I don’t know the slightly weirder stuff. Yeah. Yeah.
kegan: Once you’ve got that groundwork, then you can sort of do it
Dr. Bill Smart: Freedom of tenure, but I really started to anyone to get involved in this. And I think, you know, looking at job opportunities in the future, I think there’s gonna be a tremendous market for either lawyers who understand technology or technologists who understand legal frameworks.
kegan: Do you think that it has to be one person or you worked with a with a law professor, right?
Dr. Bill Smart: Yeah. No, I think you’d have to it’s like everything, like when you’re building a robot, you’re not both the mechanical engineer and the computer scientist, but you know, enough of the other person’s language to, to be able to collaborate with them.
I think it’s the same thing. Yeah. I don’t know the law, but I know enough of the basics of the law and enough of the terms that I can talk to my [00:17:00] colleagues who do know the law, and we can have this sort of substantive discussion about it. Yeah,
kegan: Know enough to be able to have that conversation and to be able to collaborate.
Dr. Bill Smart: Okay. I think too, you know, there’s this growing community and with, as communities grow, you, you, you start to trust each other. So now I have people who know the law, who I’m not embarrassed to go and ask a really dumb questions of. And they say, well, every first year knows that. And then they explain it and then they ask really dumb technology questions for me.
And I think that, getting past that embarrassment is a key for the sort of stuff. Yeah.
kegan: Do you have any advice for how to build that sort of network,
Dr. Bill Smart: um, come to WeRobot in September at the University of Washington, I think it’s, it’s like a lot of things you know, it’s like when you go to ICRA you listen to papers, but a lot of the, the business of the conference happens in the hallway, right. When you’re meeting people and you’re [00:18:00] talking about like slightly less fleshed out ideas, I think it’s really just building a professional network. Um, getting, you know, finding someone who’s maybe a little more, a little farther down their career path than you are, and then so asking, you know, asking to be brought to the table and so injecting yourself into those conversations.
kegan: That’s great advice. We’re getting closer to time here. So I I’d like to wrap up with, what are you most excited about moving forward? If this could be a project you’re working on, this could be competition related or just whatever you’re most excited about.
Dr. Bill Smart: I’m just really excited to be in robotics right now. I think we’re at this weird place. Um, I’ve been doing robotics now for about 30 years. Um, which seems like a long time, but I finally figured out how to build robots that people can buy and put in their homes or put in their businesses that are doing things that are practical. We can start to think about, you know, [00:19:00] how does, how do people interact with these robots for more than an hour?
For me, you know, for a couple of hours for a couple of weeks at a time or years at a time we start to, I think we start to be able to answer the more interesting questions, ’cause we know, we understand like how to make a robot go from here to here. We know how to build them up. We know these aren’t solved problems by any means, but we know enough now to ask questions that are large-scale and actually could affect how we live our everyday lives.
kegan: I, I totally agree though. So we’re finally seeing these robots come into the real world, and I think that introduces a lot of really cool, interesting new
Dr. Bill Smart: problems. And. Yeah, and I think too, like, there’s an opportunity now, if you want to get a job in robotics, you can join a company and work on a robot that in two or three years, time is going to make someone’s life measurably better. And that is really good. That is really cool. Yeah. [00:20:00] That’s awesome.
kegan: Thank you for taking the time to come talk with us. And I really enjoyed hearing all of that. You had to say. Um, so thank you.
Dr. Bill Smart: That’s my pleasure.