Embedded - 455: Snaps!

Episode Date: July 20, 2023

Natalie Friedman joins us to discuss when, where, how, and why robots should wear clothing. Natalie is a PhD candidate at Cornell Tech.  Natalie’s website is natalie-friedman.com and you can find h...er papers in the research section. She has an Instagram account: @natalie.victoria.f AIForGood shows several robots dressed in home, business and social attire. Roomba cosplaying a mouse (Instructable) Pepper is an android-ish robot made by SoftBank. There are many clothing lines devoted to dressing it for whatever occasion you need, simply search for Pepper robot clothing. What could go wrong? Natalie recommended Fashion Is Spinach by Elizabeth Hawes. It is fascinating. Transcript

Transcript
Discussion (0)
Starting point is 00:00:00 Welcome to Embedded. I am Eliseo White, alongside Christopher White. Our guest this week, well, we're going to talk about fashion and robots. I'm excited. Our guest is Natalie Friedman. Hi, Natalie. Welcome. Thank you. Hi. Could you tell us about yourself as if we met at NYC Resistor? Sure. My name is Natalie. I'm a PhD candidate at Cornell Tech, working on designing clothing for robots. So NYC Resistor is actually 20 minutes away from my university.
Starting point is 00:00:43 Cornell Tech is a satellite campus of Cornell on Roosevelt Island in Manhattan. So I guess I would say I'm an art generalist and also a human-robot interaction researcher. I studied cognitive science. So I'm looking at robots from a social science perspective. And now I'm in the field of human-robot interaction. So things that I would study are how can we make robots more socially appropriate? How close should robots physically be to you, like, say, in an elevator? And how tall should robots be to make people more comfortable?
Starting point is 00:01:19 So I'm looking at clothing for robots, and I've always been kind of crafty. I've been into jewelry and pottery, photography, painting, drawing. And it's always fun for me to learn new methods for crafting. So I wanted to learn how to sew. And this is while I was studying human-robot interaction. So I thought, like, how could robots just move more beautifully, like in animation, right? Because robots tend to be kind of rigid in their movement. So I put a skirt, I sewed up a skirt for Blossom the robot and had Blossom twirl. Okay. Short questions and we want short answers. And if we're behaving ourselves, we won't ask what exactly conductive thread do you prefer?
Starting point is 00:02:10 But maybe we will. Okay. Are you ready? Yes. Art or technology? Art. Favorite fabric? Linen. Buttons, snaps, or zippers snaps
Starting point is 00:02:26 conductive thread yes or no and if so what kind um yes and with a low resistance favorite non-fictional robot um blossom okay i'm not familiar with that and we're gonna break the rules you're gonna have to tell me gilmore girls wasn't there somebody named blossom but i don't think she was a blossom was his own show but she was not a robot she was okay who is blossom blossom is a robot made at cornell and it's made with rubber bands and wood, and it's like a DIY robot. Okay, we'll have to look up a picture of that. Sure.
Starting point is 00:03:12 And do you have a favorite place to watch whales in Santa Cruz? I would say the boardwalk. Complete one project or start a dozen? Start a dozen. If you could teach a college course, what would you want to teach? A small studio course about clothing for fashion designers or for robots. And I would teach it to fashion designers and social scientists together. Favorite fictional robot?
Starting point is 00:03:41 Eva from WALL-E. It's a popular one. Yeah. Do you have a tip everyone should know? Okay, when you buy peanut butter, all the oil is usually at the top, but if you flip
Starting point is 00:03:58 it over before you eat it, then the oil gets evenly distributed throughout the peanut butter. You store it upside down, because I have seen you put them right side up in the cupboard. I don't eat the peanut butter. I know, so why do you? So what does it matter to me? I just want to give credit to Marsha Carlin.
Starting point is 00:04:19 She gave me that tip. Okay, so robots and clothes uh you said it's it's about human human robot interaction human computer interaction it's not about clothes themselves that are robotic no smart scarves that open doors for me that yeah it's um that does sound cool but yeah i'm talking about robots that are wearing clothes um to support their own own tasks and then like indirectly your tasks, right, with the robot. So, for example, I was working on an agriculture robot tractor and I designed a tool belt for that tractor um so that's that's kind of an example right it's supporting the robot's tasks that's going up and down the greenhouse while someone grabs into its tool belt and pulls out a pair of scissors um and it's also supporting the robot oh so the tool
Starting point is 00:05:15 belt was for the human companion not for the robot right and supporting their task together okay it didn't seem like robots would like pockets as much as I do. Well, they're often like the clothing is made after the robot is made, right? So it's not necessarily programmed for an arm to reach into itself. I saw Simone Yatch did a YouTube video where she made an arm, a robotic grabby arm out of stained glass. Oh, wow. And it was very pretty, although given the tears towards the end when it failed, it seemed very difficult. you trying to make things prettier more approachable or more like the utility belt where you're you're trying to solve two problems or you're trying to solve the human's problem not
Starting point is 00:06:12 the robot's problem i would say that clothes can do a lot of those things um and i try to make it happen for each of those things so So for example, that utility belt, it was going to the Stanford educational farm and the colors that like for that farm on their logo is red. So I made sure that like there was kind of that branding on the utility belt. And then in this case, it wasn't so protective, but there actually were some holes in the robot that water could get into so it ended up by accident just covering up those holes um and then aesthetically yeah i would say that like social signaling through color is kind of an aesthetic choice um but not necessarily pretty i think it really depends on the context. So, for example, maybe Lady Gaga is doing a performance and she wants to bedazzle her mic and to match her costume, right?
Starting point is 00:07:11 So I think that that's a case where you want to think a lot about, like, aesthetics. So do you think that beautiful robots will be more alluring to humans than non-beautiful robots? I think so. I think that there's some risk making robots beautiful because they might not belong in their environment, right? Like wearing a very fancy dress to a utilitarian environment might look wrong. So I think I prioritize matching the context more than just being like aesthetically beautiful. So when should robots wear clothes?
Starting point is 00:07:58 I mean, it seems silly. I'm sorry. I know this is your PhD research, but it does seem a little silly that, I mean, I have in my mind a robot arm with a tutu on it, which is not what you're talking about. Here I am thinking this is a great idea because you get used to, okay, I know this is a little bit off topic, but it's on my mind and I'm going to forget. In an industrial situation, you get used to the machinery that's around you. And you might not see it the same way after a few weeks of working with it. Definitely. But if you were to change the appearance of a robot arm periodically, dress it up however you want with a tutu or whatever, you would see that arm and you would not get so used to it and perhaps miss something that it was doing incorrect
Starting point is 00:08:45 or decide that you didn't see it and it whacks into you. It just seems like that would be an application that would be important of changing visual appearances so that you don't get too complacent with a mechanical environment. Okay. I like that idea. It reminds me of an idea my advisor had one time about changing the clothes when there's a status update in the software. So just to remind the person, hey, this is a new and improved robot. I like that, yeah. Yeah, but in what you said, it kind of reminded me of, yeah, just changing the appearance for more attention. And it's a good excuse to dress up robots in clothes.
Starting point is 00:09:27 I mean. Right. Does it have to be human equivalent clothes or is it any covering? What are you focused on? Yeah. I tend to look at the materials and textures and colors that people would wear in the environment and match those rather than thinking, oh, it needs to wear eyelashes to look more like a human. This robot needs pants. It doesn't have legs, but it still needs pants.
Starting point is 00:09:55 It needs pants. Speaking of that, so one of the first prototypes that I made for that utility belt, I put a belt, like a long belt around the whole robot and it got stuck in the plants that like and it didn't need the belt it was just me thinking oh it needs to wear a belt because it's wearing a like it's wearing a utility belt and that is what people wear so it can be dangerous to take things, shapes and forms from people and just apply it to robots. Sure. But you do, but you are calling them clothes and not enclosures. That's a good point. This has been a really challenging topic, like figuring out what to call it. Like sometimes I want to call it fashion. But then my advisor says, hey, you're going to alienate all the engineers in a publication if you call it fashion. So clothes feels a little bit less,
Starting point is 00:10:53 I don't know what the word is. Girly? Showy, maybe? Yeah, girly, showy, yeah. That's a mundane term. It has no, I mean, clothes span the gamut from burlap sack to ball gowns. And fashion is usually high fashion. So, no, I mean, clothes span the gamut from burlap sack to ball gowns. And fashion is usually high fashion. So, okay, I can see that. Right. And then coverings is something that we've considered. And I think that's even more mundane than clothes. But like, there's a casing, right? When you say enclosure, I think of like the casing that goes around the
Starting point is 00:11:25 mechanisms in the robot. And I kind of want to, I want to think about clothes as removable and like to the average person that's not the engineer, right? So when you say enclosure, I think of, yeah, the casing that comes with the robot that is around the robot already is part of the goal with dressing robots making them more anthropomorphic um no i think i think that it's really hard to get away from anthropomorphism um but really i'm just trying to to have the robot fit better into its environment. Um, and maybe protect the robot from dangling wires getting snagged, um, or from overheating. Maybe it can insulate, um, the robot too. Um, but those are engineering functions that should be solved with enclosures. Go back to that word. I totally agree. I totally agree. But sometimes robots ends up in places that engineers don't expect.
Starting point is 00:12:32 And I think that that it happens a lot. So I think that clothes can kind of be that last step solved after the fact when we when we watch the robot in the world and see where it ends up. So you mentioned the utility belt. What other examples do you have? So things that I've just seen out in the world already, during COVID, there was karaoke mic coverings that were getting replaced. And that kind of is an example of this, right? We didn't know what was going to happen. We didn't know that COVID was going to happen, right? So that was kind of an aftermarket solution.
Starting point is 00:13:15 The mic that Lady Gaga uses during a performance. I've been thinking about thermochromic ink and the functions that it could have for robot clothing. So, for example, if part of the robot is overheating, we might be able to tell by the color changing. Yes, when the smoke comes out, it's bad. You can make it turn red and angry, like being anthropomorphic. And so your degree is an art degree or a computer degree? It's information science. So a lot of people that come out of that program tend to be user experience researchers, right? So they have
Starting point is 00:13:52 a little bit of knowledge about technology, design, and engineering. Information science used to be a degree for librarians. And it's so fascinating to me that it covers now both human-robot interaction and library sciences. Okay. I mean, those are information, yes. Yeah. How does robot clothing relate to information? That's a great question. Well, there's a lot of information that clothes carry, right? Like there's all this fashion literature that I've been assigned. So there's like higher quality materials, right? And that might carry the information that a person is of higher class, right? Because they can afford it.
Starting point is 00:14:39 And maybe that would apply to robots too. Absolutely. Think of Apple's cases versus some of the Android cases. They feel more expensive. Don't email us. That's a really good point. Yes. But, okay, so indicating the robot status, indicating the robot owner's status.
Starting point is 00:15:05 Right. Okay. That's a good the robot owner's status. Right. Okay. That's a good point. It's both. What else? What other kind of information? Yes. Um, I guess it's, I think clothing can communicate capabilities of robots too,
Starting point is 00:15:22 or maybe enhance the motion so for example you might add stripes um to a rotating robot or some kind of swirl um to to show everyone that this this is moving and kind of enhance that motion oh like on uh airplane turbines there's a little say more little there's a so on the um if you look on the next time you look at a plane at the front end of the engines there's the turbine and then there's a little, what do they call it, a nose cone or something on the turbine? And they have painted a little swirly on it.
Starting point is 00:15:56 So you can see that it's moving because when it's moving it's very hard to see the turbine blades. But you can see the swirly on the the swirly on the nose cone. That's all I got. Title option there. And you mentioned the utility belt. That does indicate that this robot is doing something related to what is in its belt. So you could indicate function. Right. And who, exactly, and who is using the robot at the time, right? So maybe when the robot is walking alongside a farmer that's cutting basil, it would have scissors
Starting point is 00:16:36 in the tool belt. But maybe when it's walking alongside an electrician who's fixing the robot or something else, it would carry tools for the electrician. What do you want to do with your career after you get a degree? That's a great question. I'm figuring it out. That's the answer that says, I don't know. I'm figuring it out. So one idea I had was thinking about like maybe starting a company or working for someone who is making clothing for people that have prosthetics. Just because I've been thinking a lot about
Starting point is 00:17:12 these questions of making clothing forms for alternative forms. Also, I have like the skills to do user experience research. So probably user experience research at some robotics company. And hopefully I can use this knowledge and smell the basil and just like interact in the field. So yeah, that's where I'm at right now. That's what I'm thinking. What led you to get a PhD? Well, as an undergrad, I really liked doing research because of the way it was just so hands-on. And I always looked up to the PhD students that got to lead the research. And my advisor, Wendy Ju, she does design research, and it is very hands-on. So I thought, wow, that would be so great to be able to do design, which is something that I love outside of school, and then also research. Where did you get your undergrad degree in?
Starting point is 00:18:28 Cognitive science. Ah, yes, you said that. I minored in cognitive science. It was really fun. Oh, really? Yeah. Awesome. That's cool. So when you say user experience, I think of my days working on children's toys where we would get a mock-up of what the toy was supposed to do and then have some parents or kids poke it at the toy. At the mock-up.
Starting point is 00:18:54 At the mock-up. And then a person in the room would do the voice that would happen for the toy. And it was incredibly adorable. The whole process was just fantastic. Is that what you mean or more large scale? That sounds like so much fun. That is what I mean. I think that what you're describing may be a little bit later stage where you're thinking about like specific features. I like to do the early stage stuff. So just figuring out like without even a research question in mind or a specific feature in mind, just like what happens when people are around this robot and they're using it as they would.
Starting point is 00:19:41 So that's usually like early stage questions where you're still thinking about like what features should exist to support the people and what they would already be doing with the robot so a robot in uniform would seem very different than a robot not in uniform and a robot in tutu um well we think about the robots we tend to see today. There's the all white kind of plasticky security robots and androids that people are trying to make the, you know, Honda or Tesla or whatever. They all have these weird white plastic forms that are unadorned or heavily industrial robots that don't look like people's conceptions of robots, right? I mean, they look like industrial robots. Yes, but... But, I mean, there are the wires and they're usually hooked around, but you can see the struts and the clamps. And the enclosures are for the little parts, not the whole big part. Right.
Starting point is 00:20:38 I would say that they're naked. I think that there's a tendency for designers to want their robots to look very sleek and to require less maintenance overall, right? If you have clothes, they could get dirty. Yes. I think it depends on the context. So in a restaurant in Japan, for example, I saw the waiters that were robots dressed as the waiters that were people. So for that context, I think it makes a lot of sense for the uniforms to be similar. They identify the robot's function in a way we understand. Right, exactly. Cool. Exactly. Did I answer your question? I don't know that I did. I don't remember what the question was.
Starting point is 00:21:28 Okay. I probably derailed it. Sorry. Do you think there is a role for having silly costumes on robots? I do. I think that they can make the robots more approachable, especially to kids. And one could imagine a performance robot, right, like being silly. I know like Simone, like you mentioned Simone Gertz, like she, her robots are totally silly, right? It's kind of the whole act. It's the intention. All right. And when I was teaching Girl Scouts how to build robots, and I brought a bunch of crafts with me, and I think that the crafting around the robot made building robots more approachable.
Starting point is 00:22:18 I think that sometimes this idea of engineering and robotics can be kind of intimidating to kids, especially girls. And yeah, adding this element of something that's familiar, making it silly, can actually make the robots more approachable and building robots more approachable. Using silliness to teach is fantastic. That is the way I much prefer to learn. I agree. How much of your thinking or not your thinking, but well, okay. How much of what you think is appropriate or what people will respond to is guided by maybe what we've been acclimated to through popular culture? Because sci-fi has robots, right?
Starting point is 00:23:06 When people think of robots, people outside of engineering maybe think of robots, they tend to think of sci-fi robots. And while they might not be clothed, sci-fi robots tend to be heavily designed for aesthetic appearance. You have C-3PO who's golden, and R2-D2 has a particular set of colors. And I think there have been robots with cape. I'm trying to think of examples of clothed robots in sci-fi, and I'm not coming up with any. I think there's been some with capes, and that's about all I can think of. Unless they're pure androids, right?
Starting point is 00:23:40 Unless they're pure androids, and then they're just acting as humans who happen to be robots, like Data or something. I mean, the Dalek, it has a bunch of ornamentation that is not clothing, but also not obviously technically useful. So I guess to distill my question, how much of your thinking is influenced by that? And how much, on a second level, how much of what you think about doing is kind of to keep people in that lane where they're expecting things i think that animation has um what's the word so animators are have have this background in character design um so they they can make the form exactly as they want right and think about the voice and the name and the yeah the textures um kind of all together um and this is kind of what inspired me to dress robots in the first place um because there there's all this character design that goes into animated characters um but that doesn't always happen for robots um
Starting point is 00:24:49 can you repeat the question i guess yeah i didn't and i didn't ask it very well it's okay uh and i think you did you did answer most of it there okay because that's you know what inspired you to get into it but um i But I guess when you're thinking about clothing for a robot, how much are you thinking about the end user and what they're expecting versus either subverting those expectations or doing something that's purely, okay, this has a purpose and it's not purely aesthetic to line up with a random observer's expectation based on their history? Yeah, I think starting with the user is usually the best place to start. So really just like observing the environment that the robot's going to end up in with the robot there.
Starting point is 00:25:44 And like watching those interactions and then coming up with the design um i think it might be different for a performance right like that that is a place where we would want to subvert expectations um i recently saw there was a ai for good um clip with a bunch of different robots um that were supposed to be representing different countries and they were different, I guess, values? And they were all wearing different clothes. I think it was Erica the robot. I have to double check.
Starting point is 00:26:14 But one of them was wearing a nurse's outfit. And one of them was wearing a totally bedazzled dress. And that was, in a way, kind of a performance, right? They're interacting with each other and that was in a way kind of a performance right it's not they're interacting with each other um and it was for film right so i think i think starting with the user when people are going to interact with it and then um kind of thinking about subverting expectations
Starting point is 00:26:38 during when you're preparing a robot for performance. And the user is the audience, the additional performer, the robot. I mean, who are you, who is the user here? Well, during a performance, I guess the audience, yes. Anyone that's preparing the robot also, right? The engineers, the hardware engineers and any additional actors, too. I have been enjoying the Murderbot diaries by Martha Wells, who we had on the show a while ago. Thank you to Ben for noting the humble bundle that had all the audio. But Secunit, who is the main character in Murderbot, uses its clothes as a way to hide. It hides behind its mask. When it later loses its helmet,
Starting point is 00:27:36 it hides behind the clothes that it uses. And while it mostly hates everything, it does eventually enjoy custom clothes. Do you think robots want to wear clothing? Do you think robots will want to wear clothing? I think if its goal is trying to hide, yes. Trying to hide its emotions. Oh, trying to hide its emotions. Okay, I see. um trying to hide its emotions oh trying to hide okay i see um i think that the word want is a tricky thing because it really anthropomorphizes the robot right it kind of assumes that this robot like has some kind of consciousness and desire um and i'm like constantly tempted to like
Starting point is 00:28:21 use language that anthropomorphizes the robot um like i want to say wire modesty or robot comfort um but right like so so if we're thinking about like i guess trying to reach homeostasis right for humans and for robots um not struggling to work, robots would want to remain comfortable in temperature and not get their cable snagged and not being squashed, right? So maybe if clothes help the robot to maintain its comfort, then yes, they would want to wear clothing okay how much of your thesis is on robotics versus psychology versus fashion um i would say most of it is psychology um and fashion um and then usually i have a paragraph at the end of that section that says, what does this mean for robots? And then I talk about how each social implication of fashion
Starting point is 00:29:36 will affect a robot's function. A lot of it, I guess, kind of the intersection of psychology and engineering is design. So I guess I would categorize the thesis as a design thesis, even though I'm formally in information science. Can you describe in layman's terms what your thesis, what the main topic is? Not to defend it here, but defend your thesis, what the main topic is? Like, not to defend it here, but defend your thesis. Gosh, are you my committee? Yeah, so the official title is Designing Clothing for an Improved Human-Robot Interaction. So I have, I guess, three main projects that I'm focusing on. Um, one, um, my design of that utility belt and like what my design process was for it. Um, the second one is, um, kind of a theoretical framework, um, for all the functions of clothing kind of together, art practitioners and researchers came up with what our predictions of clothing for robots could be. And then thirdly, I interviewed anyone that I could find that is already dressing robots. And I asked them what their design process was.
Starting point is 00:31:03 And then I'm sharing that what the current practices are compared to those predictions that I made in the previous chapter. Tell me more about the middle one about the research into clothing. So that that was the one where I was getting together with like art practitioners and material scientists and engineers to think about the functions of clothing for robots. Is that Is that the one you're talking about? Yes. Yeah, that was fun. Basically, we just had lots of design conversations about the robots our ideas into three kind of categories about the functions of closest for robots that we predicted.
Starting point is 00:31:51 And we kind of kept finding that everything, every example would fit into one of these three categories. So the first one was protection, which I've talked about. The second one was social signaling. So that means, yeah, role, individual identity. And the third one was adaptability, right? So anything that makes the robot more adaptable into its context. So like putting a jacket on a robot so it doesn't get too cold? Yeah, just like that.
Starting point is 00:32:24 Solar reflective blanket, I don't know. Yeah. Okay. That works too. Is there a corresponding theory of clothing for humans? Talk about signaling. I mean, you mentioned status and you mentioned uniform. I'm sure it's talked about, but I don't know. I've never heard of it in that concise framing before. It seems like there should be an analog. And does it fall down to the same three principles? frameworks. There's one quote that I like about fashion, and it's from a book called Fashion is Spinach. And she basically says that fashion allows you to fit in just enough and stand out just enough. So that kind of represents the social signaling aspect.
Starting point is 00:33:27 Yeah, I want to get back to you on a specific framework. Before we get to listener questions, do you have any questions for us? What's your style? Pajamas. And flip-flops. Yeah. Awesome. To be a little more serious, when I'm out and about in technical arenas, conferences,
Starting point is 00:33:58 I tend to dress more feminine than I would just going to the grocery because I can. Most of the places I go these days, I'm speaking. And so there's no, you can't really discount me because I'm a woman. And so therefore, if I dress as a woman, then people who dress as women maybe won't get discounted just for their dress. Does that make sense? It does. Definitely. So I wear skirts a lot more if I'm doing technical things than otherwise. But I also wear skirts a fair amount to the beach just because my shorts are dirty.
Starting point is 00:34:49 Awesome. Do you have a style, Christopher? I mean, other than a large, huge, monstrous collection of Star Wars t-shirts. Probably standard nerd. Yeah. Jeans, t-shirts. I've been trying to wear more non-message t-shirts. I have a lot of t-shirts that are you know star wars or band
Starting point is 00:35:05 t-shirt or some embedded t-shirts ridiculous witty thing well that's yeah i have too many embedded t-shirts because we get a lot a lot of our own shirts um but yeah i i don't i don't wear a lot of black but i think that's also standard nerd um the older i get, the less I care. But yeah, we're also beach people, so shorts and t-shirts are fairly common. You live in Aptos. But I have been starting to discover the benefits of utilitarian things like aprons. I've been doing a lot of woodworking and I was complaining to Alicia that I was getting sawdust all over my clothes and I had to keep segregating my clothes, taking
Starting point is 00:35:51 them off before I got into rooms where I don't want sawdust. I just, why don't you get an apron? I'm like, an apron? And I'm pretty sure he thought it was going to be like frilly. And I knew about them, but I went and looked, and oh, aprons with all these tool slots. So I got an apron with all these tool slots, and it's very convenient and keeps the clothes clean, and they have places for hammers and chisels and such. So yes, utilitarian clothes can be quite nice.
Starting point is 00:36:19 That's great. What about you? Do you have a style? I like to dress like a scrapbook of lots of styles. So often like bright colors and fun textures that I find at thrift stores. Yeah. Okay, I have a few listener questions. Okay. The first from Andrere from the great white north uh have you tried using the recent crop of laser cutters for textiles um i have not um but i have
Starting point is 00:36:58 used a cricket machine that hobbyists usually use um to create clothes for robots. I found that smaller parts make a lot more sense for the Cricut machine than large parts. Otherwise, you might as well just cut it just because there's a lot of replacing that has to get done. Yes. I mean, the Cricut's great, but you can only do this size or less. Right. 11.5 inches. It gets old. Right. Agreed. How much do you create the clothing yourself versus buying something,
Starting point is 00:37:38 I don't know, buying a one-year-old onesie and then modifying it? How big are these robots, do you think? I don't know. This is going to be baby size. I would say half and half. I've actually bought in doll patterns. So I made like a rain jacket for a doll. And then I put it on a keep on.
Starting point is 00:38:07 I don't know if you've seen the little yellow keep-ons um so and that was kind of an effort just to learn how to sew um it's hard to be it's hard to be a phd student right and do engineering design and then also learn how to sew like and be as good as a tailor um so sometimes i have outsourced that work uh peter wanted to know if robot clothing would be also in the form of magnetic clip-on panels that could auto detect with the robot so it understood its capabilities or so that other people could see as with what christopher was saying you maybe put on different magnetic patterns so that people don't get too accustomed to the robot. I love that idea. I would say it's, sometimes it can be risky putting magnets on robots because it can mess with the mechanisms inside. So I think there would have to be a lot of testing.
Starting point is 00:39:05 It's not as bad as you'd expect. Great. It depends on if they have Hall effect sensors. Yeah, I mean, it depends on where you put the magnets. But overall, if you're using the thin film adhesive magnets, it's not a big deal. As I've learned from my Kindle, having magnets in weird places can do confusing things. Oh, no. having magnets in weird places can do confusing things oh no um yeah i've i've put i've put magnets um on a robot and i just got a lot of sliding but i think adding adhesive to the back
Starting point is 00:39:33 would have been a great idea um there was another part to that question oh yeah so i actually have seen um robots that can sense their clothes. There was, there's a researcher at KTH, her name's Ylva Frenius, and she made a robot that could sense when it was wearing pajamas so that it would go to sleep. And I thought that was so brilliant. That seems like a lot more work than a button. I mean, it's cute, but... You got the experience of clothing a robot in pajamas. Wouldn't that get old after like the fourth day?
Starting point is 00:40:12 This is why you're not a parent. Absolutely. Where's the button, kid? Enough with these pajamas. This is very fascinating because I think as robots become more prevalent in people's lives, well, like Roombas, people do things to Roombas. I mean, that didn't come out right. But I mean, ours has googly eyes. Ours has googly eyes. Ours has googly eyes. Is that something that you've seen or is that part of your research is seeing what consumers do to consumer robots? Yes.
Starting point is 00:40:53 So I actually know someone, her name is Elizabeth Goodman, and she wrote instructables for dressing your Roomba like a mouse. And she thought, oh, the Roomba moves like a mouse, so it should dress like a mouse. So she, yeah, she has this Instructables. And I've also seen on Etsy, people were taking, were noticing that the robot was ending up under a couch and couldn't get out. So they were adding little bumpers that you could stick onto your Roomba so that it wouldn't get stuck under the robot. It's totally brilliant to me. Sure.
Starting point is 00:41:30 Roombas, the Roombas in particular, tend to form attachments. That went backwards. People tend to form attachments to their Roomba, to their specific Roomba, not to the brand. I mean, Chris Veck has told us that iRobot, sometimes they say, I don't want a different one. You can't replace it. You have to fix this one. Do you think that people will dress robots more if they feel, do you think it'll be like dogs, like little dogs that end up getting dressed whether they want to or not? I think, I think yes. We've definitely seen it with Roombas and I'm looking forward to seeing it with more. There was one YouTuber who brings Pepper along in her car and takes videos of her and Pepper dressed for the day. And she calls it Pepper Days. And one day they were going in the rain and Pepper wore a raincoat.
Starting point is 00:42:33 And I think she's probably very attached to her specific Pepper. And Pepper's the robot, not a dog, just to be clear. Yes, just to be clear, Pepper is a robot made by SoftBank. Oh. Do you think it's good for robot manufacturers to be trying to make anthropomorphic robots? Again, I'll bring up the example of Tesla, which I'm not super a fan of what they're doing there. But there have been Tesla and was it Asimo, Honda's Asimo? Is that the right?
Starting point is 00:43:06 Asimo, yeah. And there have been others where people are just trying to make, you know, the standard sci-fi Android-shaped robot. Is that a good idea? I would say no. I think that a walking robot doesn't make as much sense as a rolling robot, right? Like there are limitations that the form of our body has that just doesn't work for a robot. I think that there are also a lot of expectations that come with making a robot look like a person. For example, if you put ears on a robot, that makes people think that that's where the microphones are or that the robot can hear. And if you put eyes on a robot, it's the same thing.
Starting point is 00:43:53 So I think that there is a risk to it. But I guess, again, walking versus rolling might depend on the context that the robot will end up in. On the flip side, Boston Dynamics tends to make robots that look like evil hellhounds. Is that a good idea? Yes, because they are evil hellhounds. Or should they be dressed up in some way to make them less intimidating? Maybe a little rain jacket, a pink one with a tutu. I'm being entirely serious here. No, you're right.
Starting point is 00:44:24 Should they be less scary? I think if they're going to end up in environments where they're not intimidating people or animals, then yes, they should be less scary. There's this idea of like honest anthropomorphism, right? So you put the camera where the eyes are. So people know that someone, that the robot is watching and can see. And I think maybe that this idea could apply to kind of this intimidating movement and form factor from Boston Dynamics. Like, I guess, just be honest about what you're going to do via the form. So maybe if they are going to, I mean, hopefully they don't hurt anyone,
Starting point is 00:45:14 but if they are going to, maybe the form should match that. Sure. No. Going back to sci-fi, there's been plenty of sci-fis where friendly looking robots are actually the evil robots. I suppose they should, in that honesty, they should put the CPU in the head so that if you had to disable it with a baseball bat, you'd know where to hit it. Yeah, a secu unit doesn't keep his brains in his head.
Starting point is 00:45:37 So, do we want to trust robots? Do we want to anthropomorphize them? Is this just a bad idea? I mean, because they're not... I think what she's saying is doing it for its own sake is not a good idea.
Starting point is 00:45:54 Okay. Right? Or putting words in your mouth, sorry. No, it's okay. I think that it all comes back to responsible designers, right? So if the designers want to be honest with the users, then they should design accordingly. And sometimes that means anthropomorphizing and sometimes that doesn't. Okay.
Starting point is 00:46:19 And just don't do it thoughtlessly, I guess. Exactly. Which is pretty easy because I think, again, going back to sci-fi, a lot of designers probably start with, oh, it's a robot. Let's make it robot-y. Yeah. I mean, I still accidentally dress robots like people once in a while, right? And then I say, shoot, I should not have put that, that bracelet on the robot. Like, why did I do that? It just like makes noise and like doesn't match the context. Right. But it's so easy to say, oh, I have a bracelet. I'm just going to put it on the robot. I mean, going back to the Boston Dynamics hellhound, I don't know its name, but I think
Starting point is 00:47:00 that's appropriate. I don't know how you would put clothing on that in a way that wouldn't inhibit its movement. It looks like that for a reason. Not to intimidate, but because it needs that freedom of movement. And I don't know the specifics. Maybe you could put pink plastic all over it to make it friendly. Yeah. I mean, I think that there are options for clothing that are not necessarily fabric, right? Like you could put a decal on the robot to match wherever it's going.
Starting point is 00:47:34 I've seen that done with Relay, the robot made by Saviok. The robot goes to hotels and works at the hotel and it wears like a badge and it wears a bow tie and it wears the logo of the hotel so i think there are ways um to like to avoid inhibiting movement um with clothes without actually yeah putting fabric on the robot okay so clothes does not imply fabric no i don't think so okay i would say anything that's. And things that are changeable on a frequent basis. Is that a fair analysis? I think so. I think so.
Starting point is 00:48:13 And hopefully by anyone, right? By the user or by the engineer. Like you shouldn't have to have a screwdriver to take on and off the clothes. Okay. I have an off-the-wall question for you since we've been talking about people interacting with robots or things pretending, not pretending, but things that appear to be human or human-shaped.
Starting point is 00:48:33 How do you feel about large language models like ChatGPT and the way that they don't pretend exactly, but the way that they try to... Oh, they clearly pretend. Well, the way that their interactions with humans appear to be human-like. That's not exactly a robot, but it is a robot-adjacent thing where humans are interacting with a computer in a way that a computer is not dressed up exactly,
Starting point is 00:48:58 but is pretending or evoking human behavior or appearance? That idea of honest anthropomorphism keeps coming back to my mind. I've had the experience before where someone was using chat GPT to email me, and my trust for him significantly went down. Because we've been having conversations, and the tone was just so different. So I think that there's definitely a kind of inauthenticity that comes about with chat GPT. Were you asking about chat GPT in relationship to clothes? I'm just trying to understand your question. It's kind of an analogy that sprung in my head.
Starting point is 00:49:48 We anthropomorphize the appearance of robots, but it feels like with ChatGPT we're anthropomorphizing information or the way that a computer speaks, because the way we used to interact with computers were with queries and search results that are obviously computers or with graphical user interfaces that are obviously computers. And now moving it toward, oh, we're having a conversation with the computer and it's acting just like, well, it sounds like a human. Right. And they seem like very parallel concepts to me. I love that. I think that's a great analogy. There has been some studies on politeness, robot politeness, where a participant was interacting with a computer and then it asked that person, how did it do and people were more likely to rate it well when they interacted with that computer compared to a different one when it was talking about the one that originally
Starting point is 00:50:51 talked to so people want to look good for um or have a have a connection to a robot in a way um yeah so it kind of reminds me about like inauthenticity right so if if you say like hello i hope you had a good day to chat gbt like it'll say the same thing back and that's not real like it doesn't actually hope that you had a good day right it's just like this element of politeness that um isn't that is completely programmed and it's ingenuine because machines don't have hopes and desires, I would argue. It's a great analogy. It seems like you're thinking along the same lines. We're entering a different world here where there is a lot of inauthenticity.
Starting point is 00:51:44 And I liked your honesty approach to it, and I think that needs to be applied in more places. Thanks. But when you go away from Android robots, the authenticity, the anthropomorphic authenticity becomes more difficult. I mean, you could make something look like a bug and that would indicate where its eyes are. But then it's creepy like a bug. And I mean, I find mice kind of creepy. So I wouldn't want a Roomba mouse.
Starting point is 00:52:23 But you could talk me into a Roomba cuttlefish pretty easily. Oh, that'd be cute. The little things would help vacuum, you know, it would, the little tentacles would scoop in. I mean, it would look very Cthulhu-like, but awesome. I would love to see you do that. Yeah, I guess anthropomorphism is a bit more challenging when it's not on a human form. But, right, you put eyes on your Roomba, right? So that may signal that there are cameras in that area.
Starting point is 00:53:06 Sure, we could have put them near the camera the camera. That would have been a good idea. Yeah, it's a difficult topic. I'm glad that people are thinking about these things in this way. I sometimes feel like engineering just kind of goes, here's some stuff. We put it together. Are you going to buy it?
Starting point is 00:53:22 I have a Hall Effect sensor, an IMU, and three motors. What can I build? And then there's no thought of how do I make it good or approachable. I just have the thing. Even if I'm solving a particular problem, it's just direct. I feel like the last 10 years, design has kind of gone by the wayside and things have become more utilitarian
Starting point is 00:53:48 and flat and I think there's still a place for design and thinking about how computers and humans interact with each other not as friends necessarily definitely not as trusted friends and I'm not saying anything about my friends being untrustworthy. That was not an implication.
Starting point is 00:54:11 Natalie, it's been really great to talk to you. Do you have any thoughts you'd like to leave us with? Oh, yeah. Okay. My last thought is the best thing that you can wear is a smile. Unless you're a robot and you're not really happy. I'm talking to people right now. Or you're a robot and you don't have a mouth. I have no mouth and I cannot smile. Sorry. Our guest has been Natalie Friedman, information science PhD candidate at Cornell
Starting point is 00:54:42 Tech. Thanks, Natalie. This was fascinating. Thanks so much. Thank you to Christopher for producing and co-hosting. Thank you to our Patreon listener Slack group for questions. And thank you for listening. You can always contact us at show at embedded.fm or at the contact link on Embedded FM. And now a quote to leave you with. Let's go with Hitchhiker's Guide to the Galaxy by Douglas Adams. Don't blame you, said Marvin, and counted 597,000 million sheep before falling asleep a second later.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.