Transcript from 450: Swimming Through Nutritious Slurry with Kari Love, Chris White, and Elecia White.

Welcome to Embedded. I am Elecia White, alongside Christopher White. Compliance. It means boring, boring things for those working on medical products or devices with FCC certification. It means something far more entertaining when you are talking about robotics. Our guest this week is Kari Love.

CW (00:00:26):

Hi Kari. Welcome back.

KL (00:00:28):

Hi. Thanks for having me.

EW (00:00:30):

Could you tell us about yourself, as if we met at a Supercon or New York Resistor sort of get together?

KL (00:00:43):

Sure. I am Kari. I wear lots of hats in my life. I am really someone who makes things that are associated with fabrication and motion and softness. In my life that has been things like costumes for Broadway, film and television. As a spacesuit contractor. It is co-writing a book about soft robotics, and also teaching about soft robotics.

EW (00:01:15):

Many of those things we will be asking more about. But first we want to do lightning round, where we ask you short questions and we want short answers. And if we are behaving ourselves, we will not say, "What exactly do you mean by that?" Are you ready?

KL (00:01:28):

<laugh> I am ready.

CW (00:01:29):

Favorite fictional robot?

KL (00:01:32):

Baymax.

EW (00:01:33):

Favorite non-fictional robot?

KL (00:01:35):

Oh, it is too hard. I do not know.

CW (00:01:38):

Do robots have religion?

KL (00:01:42):

Maybe.

EW (00:01:44):

On a scale of one to ten, where one is an electric toothbrush, and ten is ChatGPT controlling a ROS 2 based chassis to roam Mars. At what point does something become a robot?

KL (00:01:57):

For me?

EW (00:01:57):

Mm-hmm. <affirmative>

KL (00:01:58):

Two or three.

CW (00:02:01):

What is your favorite medium for soft robotics?

KL (00:02:04):

Inflatables.

EW (00:02:05):

What is your favorite soft robotics drive mechanism?

KL (00:02:09):

Programmable air.

CW (00:02:11):

If you were to write another book, what would it be about?

KL (00:02:15):

It would probably be about tech plus craft, or digital fabrication with post-consumer waste.

EW (00:02:24):

Do you have a tip everyone should know?

KL (00:02:26):

If you cannot fix it, feature it.

EW (00:02:29):

CW (00:02:31):

That is like jazz. <laugh>

EW (00:02:35):

So you mentioned you wrote a book about soft robotics. Could you summarize that?

KL (00:02:41):

Sure. It was a book by Make. It is called "Soft Robotics: A DIY Introduction to Squishy, Stretchy, and Flexible Robots." And it was meant to be a maker or hacker introduction to making your own soft robots.

EW (00:02:59):

But it had a lot more theory than most of the Make books. I mean, it actually, in parts, seemed like a technical, not paper, but definitely an introduction to the whole field.

KL (00:03:16):

Oh. I think that was really intentional on our part. I think there is one thing about it that drew me to recommending that approach. One was that the majority of people do not actually make all of the DIY projects in craft books, or in maker hacker books.

EW (00:03:37):

Uhh <intake of breath>. I hope you do not find out how many I own, having never made any of the projects.

CW (00:03:40):

<laugh>

KL (00:03:41):

Exactly. Part of the joy is the aspirational, "I am going to build this." But also, I really feel like people are interested in these books, because they are interested in the topic at large. So I really recommended to my collaborator and to our editor, that we include a lot of information about why soft robotics. Because I think that, for people who have never heard of soft robotics, it is a little mysterious why.

(00:04:10):

I mean, there is the "it is cool" factor, but it is very difficult in some ways to build a soft robot. Especially for all the reasons that also make them great. That they have non-linear behavior, and surprising behavior. So to take on those challenges, you have to have a reason.

(00:04:31):

The other thing that I thought was why we included a lot of technical information, and what else was going on in the field, is because a lot of the projects in the book are things that are very similar to other things out there. We did not want it to be ripping off someone else's work, but instead it is that we are part of a discussion of the larger field of soft robotics.

(00:04:57):

So if we are building a project, there are multiple projects that have similarities with it. We are just riffing in the space. We are sharing what the baseline knowledge is, as opposed to stealing someone else's work and publishing that. So that was another reason why it is great to include what else is going on in the field, what is the history in the field.

(00:05:18):

I also just really love the idea that some of the people who read this, may not care about that at all. They will just want to see the projects. But for people who get really drawn into the idea of what soft robotics is, and also what the possibilities are, they have breadcrumbs to go and do more self-guided learning on the topic.

EW (00:05:42):

Was Make- Was your editor originally in favor of this idea, or did they push back?

KL (00:05:49):

I do not think we ever got push back on what could go in the book. I think they originally approached Matthew Borgatti, who is the collaborator on that book, about making a book that included a robot of his, that is not even in the book in the end. So that was their idea of what the book needed to be.

(00:06:10):

Ultimately, that robot did not end up in the book, because none of the technical editors, or anyone else who tried to make it, could reproduce it, because it was so difficult to make. So we ended up making much more accessible projects to put in the book.

(00:06:24):

And then we also provided all this context. I think the thing that we were passionate about was to invite people to design their own robots. And I think that that is, to me, what all of the background information invites you to do.

EW (00:06:47):

That makes a lot of sense. But you did say you do not usually do it, unless you have a need for something about soft robotics. In what case does someone need a candy based robot?

KL (00:07:00):

<laugh>

EW (00:07:01):

Other than it is awesome, which I will not deny.

KL (00:07:05):

It is so funny, because I think it is this back and forth between the idea of a blue-sky future. So if you ask someone what a candy robot is, who is an academic or working in a highly technical company, the answer they will give will be about, like, the future of healthcare robots. There are these scenarios in which we will need to be able to eat a robot.

CW (00:07:27):

<laugh>

KL (00:07:28):

And have it do something in our digestion, but not harm us. So that will be your blue sky answer.

(00:07:37):

And then there is the answer, which is right this moment, which for me is, I teach and make in an artistic context. So for me, there is an immediate artistic reason to make a candy robot, which is as far as interactive art goes, it engages all five senses and then it literally becomes part of you. And so that idea excites me.

(00:08:00):

And then I think there are all these in between the blue-sky future and the what is achievable right now and artistic intent.

EW (00:08:12):

The book was published in 2019 or end of 2018. It has been a few years. How much has changed?

KL (00:08:23):

In terms of what is available to people to work on, and what information a person at home can have? Not a lot, surprisingly. I actually am really surprised how many people who come up to me, and they are like, "Yours is the only book I have ever read about soft robotics," just because there are not any.

(00:08:43):

The other books I see about soft robotics are only engineering textbooks, that are very expensive and not very accessible. Other than online content, which not to say that online content is not amazing, there are tons of it and super useful. But if what you want is a physical book to read about entry level soft robotics, it is still kind of the only game in town.

(00:09:09):

In terms of the greater field of soft robotics, I would say the landscape for commercialization still has not grown or panned out. I would say, the same things that were looking the most marketable then, are still the most marketable now, which is grippers.

(00:09:29):

In terms of other things that are soft robotics that you would see that are commercialized, you would still see the same things you saw then, which is things that I would call a soft robot that the people making them would never call a soft robot. So that has not changed.

(00:09:47):

But I would say the research landscape has really expanded. Since 2018, 2019, and now, there are more and more laboratories and schools all over, with people doing their master's work or their PhD work on soft robotics now. And I think that that is where it has changed, rather than it changing at the market and commercialization level.

EW (00:10:19):

I often expect technologies like this that have a whizzy-ness, that is not necessarily obviously applicable to lots of things. I mean, all the things you said. But I expected to see it in toys.

KL (00:10:35):

I think you do see it in toys, but again, no one will call it a soft robot.

EW (00:10:39):

Do you mean the dancing blowy things? Would you call that a soft robot?

CW (00:10:45):

The what?

KL (00:10:45):

Do you mean the things that are advertisements in front of car dealerships?

EW (00:10:51):

Yes. That is what I mean.

CW (00:10:53):

<laugh>

KL (00:10:54):

Again, on the scale of, "call it a soft robot, to not call it a soft robot," I would call it a soft robot. But again, nobody else would. But even things like, someone was talking to me about pneumatics for cosplay. We have not really seen pneumatics for cosplay.

(00:11:11):

And I was like, "Well, we have not really seen computerized pneumatics for cosplay, but there are those fabulous hats with little bulb pumps, and then the ears flip up." If we stuck an electronic pump on that, no one would have a problem calling that a soft robot. But because we still have a hand pump, nobody calls it that.

EW (00:11:32):

The drive for soft robotics has been a barrier. I mean, it was a barrier to entry for me. I wanted to play with all the things in your book, but I did not want to end up with a bulb at the end. I wanted it to be a robot.

(00:11:44):

The winner of the 2021 Hackaday prize was in fact soft robotic drive, FlowIO. Are you familiar with them?

KL (00:11:58):

I am. I have to say there are many things that project are doing super well.

CW (00:12:04):

Can you describe it, Elecia? I do not remember that one.

EW (00:12:07):

I was going to ask her to describe it. Could you describe it, Kari? <laugh>

KL (00:12:11):

Sure. You might be able to describe it better than me, but essentially it is a miniaturized wearable pneumatic drive system, that has built-in pumps and built-in valves. It has some modularity, that some units you can add a bigger pump, for example, if you need more air. Or they have a couple of other kinds of units, that I am not remembering right off the top of my head.

(00:12:36):

I would say those are the things that they are really doing best, is they have a GUI that is really wonderful. They have miniaturization down, and they have made it wearable. They have five sets of valves, so you have a number of channels. So you can program quite complex things.

(00:12:57):

In terms of drive systems that are out, though, I would say there are always trade-offs. It is trade-offs all the way down. For me, the trade-off on FlowIO that is difficult for me is accessibility to the unit. It is a current project for someone's PhD. You cannot really get your hands on one. In order to build one, it is very, very expensive. It is probably $800 in parts. So that is a big barrier to entry. And it is also very complex to build.

(00:13:35):

Now there have been past projects that are drive systems for pneumatics. There was Pneuduino. There is the Soft Robotics Toolkit one. Those ones really suffer from not being maintained, and like have their own drawbacks.

(00:13:52):

Actually, the FlowIO people have done a good job of breaking down the different Arduino-controllable pneumatic systems, and talking about the pros and cons. So with Pneuduino, it was that it was both very expensive to get, and also it did not have the pumps on board. You had to get your air source separately from the control unit. For the Soft Robotics Toolkit, the barrier to entry was also, again, price. It was about $800 in parts to build.

(00:14:23):

The one that I teach with at my class, was actually from one of my past students, designed it for his master's thesis. Amitabh Shrivastava is the person who developed Programmable-Air.

(00:14:36):

The good thing about Programmable-Air, is that the master's thesis included crowdfunding, in addition to building and the research surrounding it. So that means compared to other units, there have been more out in the world. More than 500 units have been sold. <laugh> Which for a programmable pneumatic system is a lot.

(00:14:59):

It really wins on price point. And it is very remixable. Where that one is- So that one, the price point is about $199, and sometimes you can get an educational discount. The downside to that one, is the documentation is not necessarily where you need to be for early easy onboarding.

(00:15:22):

And I would say that is probably true of all of the available systems. The people developing them, it is hard for them to devote all the time to the hardware, and also devote all the time to onboarding new users.

(00:15:38):

Another pro about the Programmable-Air is that it is very remixable, for a person who already understands pneumatics. But again, that is maybe not obvious. Where it falls down, is the form factor. It is a little clunky. It is definitely not miniaturized. It is definitely not wearable. Although there is some ability that you can make them primary and secondary units, so you can end up with a system that is nine valves and six pumps, you still- It ends up being a very large and very clunky system.

(00:16:16):

So those are the pros and cons. I would say of the current generation things are out, you are going to be looking at FlowIO, or you are going to be looking at Programmable-Air, and then you are going to be looking at those trade-offs.

CW (00:16:32):

With traditional robotics, servo motors and things cost five cents, and are quite miniaturized now, but they were not always. I remember when I was younger, doing radio control stuff, servos were large and clunky.

(00:16:44):

Is that a matter of inherent complexity of the pneumatic systems, that they are not there yet? Or is it just it is still early, and the research and industrialization has not happened as much as it has in traditional controls?

KL (00:16:59):

Yeah, I am not really sure. Unfortunately, this kind of hardware is not where my strong suit is. My strong suit is in the compliant part of it.

CW (00:17:09):

Gotcha.

KL (00:17:09):

The soft part, the materiality of it. So the nitty gritty of how we are going to be costing pumps and valves, is not where my layer of expertise is.

EW (00:17:20):

Fair enough. There are a lot of questions there, because as you say, it is quite expensive. It is quite complicated, and there is not the equivalent of a DC motor that you can just try out. Maybe someday.

KL (00:17:37):

I think one of the things that is interesting about teaching with Programmable-Air in my class, is that it seems like having a kit like Programmable-Air or FlowIO is really important for people to be able to envision what pneumatics can do. But once they actually understand what pneumatics can do, a lot of them opt to program their own thing.

EW (00:18:00):

Well, that is kind of disappointing <laugh>.

KL (00:18:02):

<laugh> How is that disappointing?

EW (00:18:07):

Sometimes they just want it to work. I do not need another pet. I just need a tool that I can skip to the end, and do my application.

KL (00:18:18):

Yeah, that makes sense. I think it is mostly that, I would say, in terms of what the capability people are trying to get, that maybe they are struggling with, is either they want to do it cheaper. So that is one. They want to do it smaller. That is the trade-off between miniaturization versus not miniaturization. They want to make it wearable. And they want to be able to make it either high pressure or high flow rate and have vacuum.

(00:18:52):

So again, because of the complexities of the pneumatics, you can end up with all different ways that you could change what you are doing. Maybe you want to use compressed air, instead of using pumps. Maybe you want to use a larger pump. Maybe you want to daisy-chain your pumps, so you are getting higher pressure. Maybe you want to higher flow rate. There are all of these things that people end up deciding they need, that it ends up being that pneumatics is not a one size fits all solution.

EW (00:19:25):

I suppose that makes sense. And I am really glad that there are some more kits to try out, because yes, as you said, with the FlowIO, it is kits and they are expensive. You can buy ten kits and it is less expensive, but not that much less expensive <laugh>. I mean, it is good if you are teaching a course. You teach a course. It is at NYU where you teach, right?

KL (00:19:51):

Yes.

EW (00:19:54):

Could you come out to California and teach one?

CW (00:19:55):

<laugh>

KL (00:19:57):

If someone wants to fly me and pay me, I am happy to teach.

EW (00:20:02):

<laugh> How long have you been at NYU?

KL (00:20:05):

I have been at NYU since I think 2016. It has been a long time.

EW (00:20:12):

You teach another course there, which I came across as I was looking at what we should talk about. It is called "Considering Religious Robots." I have so many questions like, "Are the robots religious, or are we worshiping the robots?" Is it considering religious robots, or considering religious robots? What is this class about?

KL (00:20:40):

The genesis of this class was twofold. One was that I was doing research with a PhD candidate at Cornell Tech named Natalie Friedman, who is doing her work on robots wearing clothes, and why would a robot wear clothes, and can we add capabilities to robots by having them wear clothes?

(00:21:02):

I was bringing her lots of examples of robots wearing clothes in the world, and talking about them. One of the things I kept noticing was that you would see robots wearing religious vestments. It was a recurring theme. I was very interested in like, "Why is this a recurring theme?"

(00:21:20):

So I dug deeper and saw that there were lots of robots that were built or working in religious contexts, that were not really being considered. So that is where the "considering" comes into this.

(00:21:36):

The next stage of what happened was that the pandemic happened. We switched to online teaching, and soft robotics was not something the department felt I could teach online, because it is so fabrication heavy. So they said, "Is there anything else you would be interested in teaching, while we are in online only university?" And I said, "Well, I have a couple of ideas, and one of them is this religious robots topic." They were like, "This sounds fascinating. We want you to do something with it."

(00:22:05):

What I ended up doing was looking at all of the existing robots working in religious contexts, and really using that to ground where the conversation would go. I have found to date about 45 robots that are working in religious contexts, but I look forward to finding more that I have not yet uncovered.

(00:22:30):

The way I structured the course was that I built it as a co-learning class. So I got a lot of people who were experts on one aspect or another to give talks to us. I would facilitate finding the people we would hear from, and then have discussions about what we learned. Also the students would bring topics that came up for them, and they would guide the discussions as well. So that ended up being super broad.

(00:22:59):

We are talking about things like religious work as labor. About caregiving. We are talking about religious tools, as a way to make you be more observant. We are talking about doing religious rites for robots, that certainly happens in the world. We talk about things like, "What about digital afterlife?" Or, "How is religion effective if we decide to become transhuman cyborgs?" So it is kind of every aspect of where religion is touching against technology and robotics.

EW (00:23:36):

Could you give examples of robots working in religious contexts?

KL (00:23:41):

Some of them are quite mundane. For example, during a swine flu outbreak, someone built what is essentially like a soap dispenser in a bathroom, that can dispense holy water. Because where it was, they were worried about waterborne illness, and they wanted a way for people still to be able to get their own holy water, which was something that was common in the churches there. So, essentially you put a little thing under it, and can capture your own holy water from a closed system, instead of an open system.

(00:24:16):

Another example would be, there is a Buddhist robot in China that is a chatbot. It rolls up to you, and if you talk to it, it will answer some religious questions for you. But if you ask it a question it does not know, it will tell you that it is a very stupid robot, and that you should speak to a monk instead.

(00:24:40):

Again, I am really into things that people would never call a robot. So one of the things that I would say is, there is a lot of technology that is to- Like, Shabbos elevators. They stop on every floor, so that the fact that you are not allowed to press an elevator button if you are Orthodox Jewish, is accounted for in public settings.

EW (00:25:05):

All right. Yeah, that all makes sense. I was kind of, "What in the world can you mean?" But those all, yeah, okay. I can see how that would be useful. Yes.

CW (00:25:15):

Do you touch on what religions say about technology? Kind of the flip side, because it does not seem like something that is explicit really in a lot of religions being-

EW (00:25:28):

Thou shalt not robot?

CW (00:25:29):

Thousands of years olds. A lot of them.

EW (00:25:31):

I mean, there are golems, so it is not like there is-

CW (00:25:33):

Sure. Yep. Okay. That is a good example.

KL (00:25:37):

One of the questions that comes up is, "Can a robot join a religion?" Golems is a great example. Pretty much every religion is not a monolith. So if you ask the question like, "Can a robot convert to Judaism?" That is a question that is asked.

(00:25:54):

Some people, some rabbis come down on "yes," and some come down on "no." Some of the people who are coming down on "yes," are like, "Well, there is some discussion whether or not a golem can be part of minyan." So they use existing religious debates to then inform about whether the role of robotics-

(00:26:19):

Different religions are more accepting and less accepting of robots. Like in Islam, there is definitely a much higher occurrence of people really not wanting humanoid robots. Some of the discussion is, it is tied to the prohibition of having depictions of the human form within the mosque.

(00:26:49):

Then there are some cases where you have exceptions, for example, of things like dolls for children, where the argument is that a doll for a child serves a functional purpose of teaching them about caregiving. So therefore it is an allowable depiction of the human form.

(00:27:07):

And again, even in Islam, there are some where the prohibition is only in the mosque, and other sects are like, "Humans should never be depicting other humans." I think that is really interesting.

(00:27:21):

Even, I would say, religions that before did not have prohibitions. So for example, it was pretty common, as a fad in India for a while to have aarti, to have a robotic arm do it. And then over- It was not very controversial. It was like a fad and it was happening everywhere. But I would say within the last couple years, there has been a lot more pushback, of saying like, "Wait, is this somehow sacrilegious? Is this somehow problematic?"

EW (00:27:53):

Wait, wait. What are they adding an arm to?

KL (00:27:57):

There is a ritual for Ganesh, where you wave a candle or something else around the statue of Ganesh. It became a fad for an industrial robot arm to do this ritual.

CW (00:28:12):

I guess it depends on how Ganesh feels about it.

EW (00:28:13):

On the one hand, that is very cool. On the other hand, that does seem like if you are supposed to do it, you should do it.

CW (00:28:20):

Automating religion.

EW (00:28:21):

Automating religion seems a shortcut, that perhaps- I do not know. This, yeah. Okay.

KL (00:28:28):

But again, it is all depending on your viewpoint. Tibetan Buddhism has had prayer flags and prayer wheels for centuries.

CW (00:28:38):

Right.

EW (00:28:39):

And I was going to ask you about religious rites for robots, thinking that was just beyond my comprehension. But I do believe that if our iRobot vacuum cleaner died, I would...

CW (00:28:55):

You would what? Give it a Norse funeral?

EW (00:28:57):

<laugh> Yes. That little sucker annoys me. <laugh>

KL (00:29:04):

You make jokes, but actually there was a very famous instance, where they had a mass funeral for AIBOs in Japan.

CW (00:29:10):

Huh-oh. Yeah.

EW (00:29:10):

Oh yeah. That was sad. And I know people get really attached to their home robots, like the vacuum cleaners. Wow, that is a fascinating topic.

CW (00:29:26):

It sounds like something- The initial impression is it sounds like, "What are you talking about? That does not make any sense." But if we do not start thinking about this stuff now, when robots are even bigger parts of people's lives, people will not have thought about all of these implications, and how all this stuff works. It does go to stuff like you are saying, like people's really personal relationships with robots, which is going to extend to religion.

EW (00:29:53):

I was thinking as a comparative religious class. Applied comparative religion.

CW (00:29:58):

Yeah <laugh>.

KL (00:30:01):

I have to share this tidbit though, while we are talking about it, because I am like, "I do not know where this slots in this conversation," but I am so excited about the detail. I mentioned just in passing, what happens if we become cyborgs, to our religious life?

EW (00:30:14):

Mm-hmm. <affirmative>

KL (00:30:16):

So there are a bunch of transhumanist organizations, that are also tied to religion. So there are Christian transhumanist groups, and Buddhist transhumanist group. But the largest religious transhumanist group is specifically the Mormon one.

(00:30:31):

There are religious reasons why that is the case. Specifically that there is a mandate to make yourself more Godlike. And so the people who are adherence to Mormon transhumanism, are under the belief that if I am supposed to make myself more Godlike, I should be using any means at my disposal to do that, including technology.

EW (00:30:59):

Hmm. All right. I am glad that we can have different opinions, because it will lead us in many interesting directions. I am not sure that I want some of the transhumanist stuff, but maybe that is just because-

CW (00:31:15):

What if you had filters in your eyes, that blocked out all blinking lights?

EW (00:31:17):

I would love that.

CW (00:31:18):

See?

EW (00:31:19):

Oh so good. But I could just wear an eye mask, then I can block out everything.

CW (00:31:25):

That would cause other problems.

KL (00:31:28):

Trade-offs all the way down.

EW (00:31:29):

<laugh> Exactly. We have a couple of listener questions, that are mostly about soft robotics, because that is what I asked. Are you ready for those, or do you want to continue? I mean, a part of me just wants to talk about religion and robots, because there are so many fascinating things. What about you Kari? Do you want to go on, or stay with religion?

KL (00:31:53):

It is funny, because since you sent me an outline ahead, I prepped about soft robotics <laugh>, and I did not prep about religious robots, but I am happy to talk about either.

EW (00:32:05):

Well, let us get some listener questions in, while Chris and I formulate some other religious ones. Bailey asked, "How realistic or unrealistic is Baymax?" And if anybody does not know who Baymax is, it is a character from "Big Hero 6," which is an animated film. As well as, there are some add-ons with TV shows. It is a large helper robot, focused on medical things.

CW (00:32:38):

And it is kind of like made of balloons.

EW (00:32:40):

It is inflatable. It is made of balloons. At one point it gets a cut, and it uses tape to fix itself. It is also very sentient, but that is what happens in animation.

KL (00:32:55):

Yeah, I think that when you ask how realistic Baymax is or is not, it depends what you mean by realistic. So all of the robots in "Big Hero 6," the animated film, are actually based on ideas of robotics research that was current at that time. So collaborative swarm robots, and soft healthcare robots, were things that appeared in the movie, that they used specifically because people were researching these topics.

(00:33:26):

At least with roboticists I know, there are definitely people who are working on healthcare robotics. There are people who- Especially things like feeding people, or being able to turn people, and be able to sense if they are going to have bedsores or other problems from not being moved. People are definitely working on those problems. So I do not think Baymax is unrealistic in that sense, that people are going to continue to try to develop healthcare robotics.

(00:33:55):

On the other hand, what is unrealistic about Baymax, is that Baymax is such a multipurpose robot. I think we are going to see unitaskers for the most part, or things that have very specific applications. Because Baymax is like very widely applicable in terms of the things he can do. He is a mental healthcare robot, in addition to being able to be a first aid robot.

EW (00:34:21):

And he is human shaped generally.

KL (00:34:24):

Yes.

EW (00:34:24):

Which for a robot that could monitor someone who is confined to bed, and be able to move and help them, there is no reason why they need to be human shaped.

KL (00:34:39):

Exactly.

EW (00:34:41):

What about the mental health aspect of it? There are robots that are supposed to help you with, let us say, remaining positive in negative situations.

KL (00:34:53):

Yeah, I think that part of it is very realistic. Especially in terms of companionship robots, if you look at something like PARO, the seal robot, PARO is a mental health robot.

EW (00:35:07):

PARO is great. Okay, William asks, "What is the definition of a soft or hard robot? Can you have blended ones?"

KL (00:35:17):

I definitely think you can have blended soft and hard robots. Very few soft robots are actually a hundred percent soft. Most of them have some hard components still, these days. I, in general, am not a purist. I am like, "Yeah, mash it up. Whatever works." But I usually define soft robotics as, "robots that get their primary function and capabilities from soft components."

(00:35:43):

And then to roll that all back one step further, I find it useful to define "softness," because "soft" means a lot of things in the English language that have nothing to do with each other. When I talk about softness and robots, I am usually referring to flexibility, compressive elasticity, or tensile elasticity. So that comes down to can it bend, and then bend back to another shape? Can you squish it, and have it returned to another shape? Or can you stretch it, and have it returned back to its original shape?

EW (00:36:23):

So you have the soft robotics being part of the mechanism. So if I have a gripper arm, that is just squishy plastic on the end where the grippy part is, that does not count?

KL (00:36:39):

To me, I would not count that. Another example that I am like- People may or may not say, "This is a soft robot." For me, a lot of sex robots I would not consider a soft robot. Because the majority of their functionality is coming from the hard components underneath, that are just covered in a humanoid layer, that is squishy.

EW (00:37:00):

But if I had my gripper blow up into the shape of a claw, and be able to surround whatever it wants to pick up, that is more of a soft robotic. Even if the rest of the arm is hard, that part that encases what it needs through pneumatics, is the soft robot part.

KL (00:37:20):

Yes, that is how I personally would define "soft robotics." We have talked a lot about pneumatics, because I feel like pneumatics is a good entry point for understanding soft robotics, but I definitely also do not want to say, "Pneumatics is the end all and be all of soft robotics."

EW (00:37:38):

You mentioned stretchability and flexibility and squishiness, but setting aside squishiness, the other two I can do with paper and flexible circuits. At what point does that become soft robotics?

KL (00:38:01):

I would say, if you are using the paper specifically for its flexible properties, that it could be a soft robot.

EW (00:38:11):

We had Carl on the show recently, Carl Bugeja. He does this thing, flexar, with a flap actuator. He uses a magnet on one side, and what looks like an antenna on the other, or a speaker wire. And he can make it flap back and forth. I sent you the video, did you get it?

KL (00:38:36):

I did. I saw it. It is amazing. I want to play with it.

EW (00:38:40):

Is that a soft robot component?

KL (00:38:44):

I would call it a soft robotic component, but I do not know if Carl would.

EW (00:38:49):

Well let us call him up and find- Oh wait, no, it is like 3:00 a.m. in his-

CW (00:38:52):

Surprise! Sweet, Carl has been waiting this whole time to answer this question.

KL (00:38:55):

<laugh>

EW (00:38:57):

Have you seen other technologies like that? I understand the pneumatic stuff, but the other flexible mechanisms like that?

KL (00:39:10):

Oh yes, definitely. I ended up, in my class at least, I give them long lists at the beginning. I am like, "Things that have been used to make soft robots," and then I have a long list of materials. Then I am like, "Ways people have actuated their soft robots," and then I have a long list of ways people have actuated it. Honestly, there are so many ways.

(00:39:35):

That is one of the things that I think is most interesting about soft robotics, as well, is that lots of times there are robots that have nothing in common with each other, except for their material compliance. Or there are things that will have multiple forms of different drivers at the same time. So you will have one that is both photonic and magnetic in the same robot, working together.

EW (00:40:02):

Photonic?

KL (00:40:04):

Actuated by light.

EW (00:40:06):

Huh!

KL (00:40:07):

Essentially, if you take a soft material, and then a way that you can manipulate that soft material, those are pretty much the components of how you can make a soft robot. So it is like, "Oh, is this actuated with-" There are plenty that are actuated with a magnet. There are some that use things like nitinol, where it is like, "Okay, you heat it up with electricity, and then it flips."

(00:40:28):

So I think that maybe people's conception of what a soft robot is, is more centered on things that are pneumatic. But I am very excited about the non-linear behavior of other things.

(00:40:44):

One example I would give, is that in my class this year, I wanted to do a co-learning experience. So we made our own conductive polymers around soft materials. We made a big dye bath of polymer. Everyone brought their own soft samples. We threw them all in a vat and mixed them up. When they came out, we tested their electrical properties, and then people applied them to different problems.

(00:41:09):

One of the ones that turned out beautifully, was they had made conductive sponge. And because of all of the non-linear behaviors of a sponge, it was very possible for them to detect different kinds of gestures. A slap was different than a poke, was different than a squeeze, because a sponge is so non-linear in its behavior.

EW (00:41:37):

I do not understand. They brought things and threw them in a pile? What? Was this resin?

KL (00:41:47):

It is a project actually you can see on Hackaday. It is called PolySense. I was introduced to it by Cedric Honnet, who is one of the people who reversed engineered a product that they were interested in, and came up with this very simple method for making your own conductive polymers. Functionally, the way it works is simply like it being a dye bath, but instead of imparting color, it is wrapping the polymers into the fiber as it forms.

EW (00:42:19):

Okay. I have opened the page, but I am not going to look at it, because otherwise I will just get very distracted. Okay, so that would make your own sensor, as well as actuator?

KL (00:42:33):

Yes. That is central to the way I am approaching soft robotics. So the thing is, my class is called "Exploring Concepts from Soft Robotics," as opposed to something like "Intro to Soft Robotics." That is because for me, my goal- Again, when I was talking about the book, it was very similar.

(00:42:53):

It is that I want people to look at materials and say, "What if I use this material in a way it was never intended to? Can I make something totally novel and exciting, by reimagining what is possible, because of a material's inherent properties?"

EW (00:43:14):

Fascinating. We may have to invite you back very soon, after I have played with this some. You mentioned the sponge. Were there other interesting things that happened?

KL (00:43:26):

That was the one that the students took it the furthest. They decided to do their final project, based on that initial experiment that we did.

(00:43:32):

Someone else ended up accidentally making a silicone hand conductive, which was not something that we thought would be able to happen. But they had made a glove conductive, and then while you are working with this material, sometimes it sheds little dust of this polymer. And so they had shoved a silicone hand cast inside the glove, which then coated it entirely in this conductive powder. And then they could use that as a sensor, which was a total surprise.

(00:44:05):

Another thing that happened was that someone decided to dye flowers, artificial flowers. They took the polymer beautifully, and they could be things like capacitive sensors. So you could make a whole thing of flowers you could be near, to cause something to actuate or to light up or whatever electrical purposes you would want to use them for.

(00:44:33):

One of the things I threw in was a pine cone, which I found a surprising thing about pine cones by doing that, which is pine cones close up when they get wet again. And I was like, "Oh, I could design a very slow interaction, if I time-lapse filmed it, by embedding things inside the pine cone, and then soaking it in water, having it close, and then have it open very slowly."

EW (00:45:01):

I could add some origami to the next bath. I have some that is pretty water resistant.

KL (00:45:06):

Mm-hmm. <affirmative> But it has to also be absorbent.

EW (00:45:10):

Oh yeah. Somewhere on the edge of- Yeah. Okay. Now that I am full of ideas, let me get back to the listener questions. Is there a simulation software for software robotics? From Uwe.

KL (00:45:28):

So the answer is "kind of." Which is, again, all these questions are always like, "Maybe. Depends what you mean." So the thing is about simulating soft robots, is that it is much easier to simulate things that are flexible, than it is to simulate things that are compressed or stretched.

(00:45:49):

So if you wanted to look for things that like flat patterned inflatable style stuff, or fashion based stuff, there will be models that you can use that will be able to predict the behavior of those things.

(00:46:04):

What happens though, when you start to get into this stretch and compress area, is that the number of variables go up in terms of the behavior, and they are harder and harder to characterize. One of the reasons they are harder to characterize, is because the tiny details in terms of manufacturing tolerance, will become much more pronounced as you get into these complex behaviors. Another thing that happens is, as these things cycle over time, their behavior also changes.

EW (00:46:44):

Which makes it hard for a production application.

CW (00:46:46):

Mm-hmm. <affirmative>

KL (00:46:48):

Yes. But it is also in terms of the predictive nature. In terms of things that are stretchy, maybe there are proprietary- Like making car parts that are out of galvanized rubbers, that are specifically very precisely manufactured, where they could predict the behavior.

(00:47:09):

But if you are doing, like even DIY in an academic setting, if there is a gap between your manufacturing capability and your modeling capability, it will often be a surprise what the behavior is.

(00:47:27):

So when you see models of things that squish and stretch, that have modeled capabilities in terms of how they are being controlled, like controlled feedback, a lot of times what happens is, they may have developed a model before they made it. They made it. It did not behave precisely like the model ahead of time. It has a different behavior. And then they feed back in the real behavior into the system. If that makes sense.

EW (00:47:59):

Totally. But do you write your own simulation software, or is this just experimental?

KL (00:48:07):

I do not work in simulation. But I think most people are using, for example- What is the name of it? Off the top of my head, I am drawing a blank at the moment. But there are a couple of different academic models, that people use that you can feed your system into after the fact, then turn it into a control model.

EW (00:48:27):

Okay. I may ask you for those later. I will put them in the show notes.

KL (00:48:30):

Perfect.

EW (00:48:32):

Let us see. From Sila, "How is power handling and delivery, different from normal robotics?"

KL (00:48:43):

So power handling and- I am sorry, what? Could you restate that?

EW (00:48:49):

Power handling and power delivery. How is that different from normal robotics?

KL (00:48:58):

Again, it will depend on what your system is. You know, if you are working with electromagnets, versus working on light. A lot of light controlled things nowadays are getting more complex, which is quite interesting, because now you have wireless LEDs. And now that you have wireless LEDs, your control system can be totally offboarded from your soft robot, which is really exciting.

(00:49:20):

Again, if you think about the domains in which you could, how you can control a soft robot, that answer is as complex as what it is you are doing. Is this a cable control? Is this thermally actuated? Is it magnetic? Is it a chemical reaction? So that question does not really apply equally, across the domains of ways to drive a soft robot.

EW (00:49:50):

Have you played with the wireless LEDs?

KL (00:49:53):

I have not, but it is definitely on my "to try" list now.

EW (00:49:57):

I know. So if anybody has not seen these, Adafruit has a small inductive coil and ten wireless LEDs for 15 bucks. And the LEDs look like- They are just little LEDs alone. They are tiny little marbles, not really round, but sort of. Maybe a third or a quarter the size of a normal regular marble.

CW (00:50:27):

Okay. So like large beebees.

EW (00:50:28):

Large beebees. Okay, I will go with that. You put them near the coil and they light up. If you think about that the coil could be whatever you want, and the lights could be embedded in whatever you want. Now you have a stuffed animal that lights up whenever it gets put to bed, or something like that. It is just-

CW (00:50:52):

Without batteries or any other electronics in it.

EW (00:50:54):

Right. Without the electronics in the device. Instead, the electronics are part of that coil. Yeah. Are there- Wireless seems like the way to go, but I do not think we can live in inductive coils. That seems harmful?

CW (00:51:12):

Fine.

EW (00:51:12):

<laugh> It is all fine. Are there other soft robotics things, that are kind of like that?

KL (00:51:21):

In terms of?

EW (00:51:23):

In terms of magicalness.

CW (00:51:25):

<laugh>

KL (00:51:26):

<laugh> Biological robots to me are like pure magic.

EW (00:51:31):

Indeed.

KL (00:51:33):

You have to feed them.

EW (00:51:34):

<laugh>

KL (00:51:34):

They eat the thing they swim through.

EW (00:51:39):

Ah, okay. What biological robots are you talking about in particular?

KL (00:51:45):

There is one that is a ray that is made out of rat heart cells, that is wonderful. And then they made a second generation one, that is in a different sea animal shape, but it is alluding me at the moment. But if you look at "rat heart cell ray," you can then see their newer work at the same time. To me, it really is magical, just see these cellular things swimming through some kind of nutritious slurry.

EW (00:52:15):

We talked a lot about that on the last time you were on the show, which was like 200 episodes ago. That is a long time. It is a good thing I get to talk to you more often than that.

KL (00:52:27):

<laugh>

EW (00:52:30):

Could a DIYer build a simple soft robot? Maybe something that could give a foot massage?

KL (00:52:40):

So a foot massage is an interesting case, because when I read that question I was thinking, "Oh, a lot of times, massage devices actually rely on their hardness very intensely."

EW (00:52:51):

Yeah.

KL (00:52:51):

So things like Theraguns are literally hard thumping devices <laugh>. The kind of massage you could get out of a soft thing, would be very different than the kind of massage you would get out of a hard thing. So it depends on the sensation that you want to get.

(00:53:07):

When I think of a soft robotic massager, I am like, "Well, you have the realm of compression as a possible feeling." You could have- I think what might be truly novel about a soft robotic massager, is if the kind of massage you like is a tickly feeling. Maybe that is something soft robotics would be uniquely positioned to do.

(00:53:29):

But I do think that DIYers can build simple soft robotic concepts. I guess I am biased, because that is what I do, is I teach people who do not know anything about soft robotics, how to build their simple ideal. So I do think that that is quite possible.

EW (00:53:52):

What is a simple soft robotic thing that someone could do to get their feet wet? Assuming they understand traditional robotics?

KL (00:54:00):

I think that probably the classic example is going to be a gripper of some kind. You are going to make either a positive pressure gripper that goes around something, or you are going to make a vacuum gripper that is a jamming gripper. That is very popular in terms of the world of possible marketable robots.

(00:54:18):

But it also is just a good way to get started. Partially because there are so many versions of them. So if you google "DIY jamming gripper," you are going to see people who are doing all different versions of that, and you can pick the one that appeals to you, that works best with the tooling that you have. Similarly you can do that with a positive pressure gripper.

(00:54:41):

So you can see there are versions of that where it is on Instructables, and you are just using a cardboard mold and wrapping it with ribbon for external restraints. Or there are some versions where you 3D print some beautiful mold. There are ones that you cast it in parts, with laser cut pieces. So you can find a version of one of those projects. That will definitely be a good entry point, if you want a guide, and will match the tooling that you have.

(00:55:17):

I think that you already identified, and we spent a lot of time talking about dissatisfaction of using a hand pump, versus using something that is computer controlled. So for me, I do not mind prototyping with just hand pumps, and other simple ways of controlling air.

(00:55:36):

But I think the most important thing that you probably need to think about, in terms of robotizing this, is do you want it to be vacuum and ambient, if you are making a jamming gripper? Or do you want it to be- If you are making a positive pressure grip gripper, is it positive and ambient? Is it positive and vacuum? That will definitely change the speed at which your actuation happens.

EW (00:56:06):

And when you say positive, that is you are holding it when it is in vacuum, or you are holding it when it is under pressure?

KL (00:56:15):

Yes.

EW (00:56:16):

And then negative would be, you are releasing it under that condition?

KL (00:56:21):

Yes.

EW (00:56:23):

Yeah. Let us say you do not want to play with air. Are there other? Can I still do a gripper of some kind?

KL (00:56:36):

You can do cable control things. That is another good thing that you could try out. You could try out something where you have made something that is a flexible part, and when you pull on a cable, then it causes that part to flex, to grip around something. So that is another way you could make a gripper, if a gripper was the thing you wanted to do.

EW (00:56:55):

Oh, I think the last time I talked to you, we talked about straws, and I made little straw fingers by cutting in certain areas.

KL (00:57:12):

Yes.

EW (00:57:13):

And then you can use a motor to roll up the string, and then the little straw fingers-

CW (00:57:22):

Oh, okay. Yeah. Like almost puppet...

EW (00:57:25):

Yeah, those puppety fingers.

CW (00:57:26):

Yeah.

KL (00:57:28):

Yeah. That is the easiest compliant system to make, that has that function. But, there are all kinds of things too. Like if we are talking about people who are already working in the digital fabrication space, there is a lot that you could do with flexible filaments, and print and place kinds of things, that would also have that same function, but maybe be more durable, or more aesthetically pleasing than a straw.

EW (00:57:55):

Oh, yeah. It is basically you have a tube, and you can do things interestingly with a tube that can bend. It is just that straws are cheap.

KL (00:58:05):

And straws are easy.

EW (00:58:06):

<laugh> Hanging around.

KL (00:58:09):

Though weirdly straws are getting harder, at least in New York City. I used to send my students to McDonald's, because they had great straws that were for milkshakes, and they made lovely, lovely bending things. But then this year I sent them, and they were like, "They have converted all to paper straws." And I was like, "Oh no, I need to figure out who is using very eco unfriendly straws." <laugh>

EW (00:58:32):

Let us see. Christopher, do you have any more questions?

CW (00:58:34):

Yeah. I think, briefly, we barely touched on it, but is 3D printing something that is getting more useful in software robotics? Because I know there are lots of different filament types that are squishy or flexible now. But they seem to have some issues that regular printing does not.

KL (00:58:53):

I would say definitely. I would say that it was already very useful, if you like working in silicone to make molds. So that is one thing. Another thing is that you can print in flexible things, either with NinjaFlex style things, or by using resin printers that have flexible filaments.

(00:59:13):

And then I also, because I have the resources of a university around me, I have been making up workflows to use other things. So for example, doing high heat resin parts, that then I vacuum form the molds. You can play around with the digital fabrication tools, to do all different ways of working with these materials.

(00:59:36):

One that I am very interested in, but have not done yet, is essentially you can make molds using a resin printer, where you make the mold itself eggshell thin. And so then you no longer have to worry about undercuts, because you just crumble the mold away when you are done.

CW (01:00:00):

<laugh> Oh, okay.

KL (01:00:02):

And I am very excited about this, and have not tried it yet.

CW (01:00:04):

Huh.

EW (01:00:06):

I do not do a lot of 3D printing. What does that mean, Christopher?

CW (01:00:10):

It means where you have got areas that you have got difficulty- I am trying.

EW (01:00:15):

The overhang.

CW (01:00:16):

Yeah.

EW (01:00:17):

No, that part I understand, but...

CW (01:00:19):

It is easier to deal with afterward, because then the mold just disintegrates. You no longer have-

EW (01:00:24):

Oh, oh. Yeah. Okay.

KL (01:00:26):

So to use an opposite example, probably more people are familiar with things like lost wax mold, where you melt away something from the inside of something else. But this is the opposite that you do. The outer part of the mold is intentionally disposed of and broken.

CW (01:00:47):

You peel it away.

EW (01:00:48):

Yeah, I know, I got that. I finally got that.

CW (01:00:50):

No, I am imagining it myself.

EW (01:00:53):

I feel like some of the reasons I do not think I can do soft robotics is my own creativity. I have always put it off, because pneumatics is hard, pneumatics is expensive, loud, big, just not an area I am super excited about, despite the fact that I have 15 pumps on my desk upstairs.

CW (01:01:19):

I was about to mention that your desk is currently covered in the pneumatic system.

EW (01:01:25):

But between strings and bendy things, and being able to build molds, I think really it is me limiting me, not the technology.

KL (01:01:40):

I think that one of the things I definitely face with students, is that people can be really scared to get started, because it does seem so daunting. It seems so daunting to not have a tutorial to follow, to have to create your own path.

(01:02:00):

For some people that is very exciting, and that is where they want to be. And then for other people, they want to be there, but they feel inhibited. I do think that a big part of what I do, is encouraging and supporting, just to say like, "Hey, it is okay if the thing you want to build, is a thing that you do not know how to build yet. We know there is the path of failure along iteration, but you will find it so rewarding, the first time you build something and it works."

(01:02:37):

And I definitely went through that this semester with someone who came to my class saying, "I want to make architectural scale inflatables." And I was like, "Great, this is a thing we can do together. This is a thing I can guide you. We have the tools, we have the fans, we have the material. Let us do it." And then-

EW (01:02:55):

Is it a bouncy house?

KL (01:02:57):

Yeah. Like a bouncy house. But more with different aesthetics, than Mickey.

EW (01:03:03):

That is fine. Yeah <laugh>.

KL (01:03:05):

Until they got started, every week they were like, "I did not feel ready," and, "I do not know how to start." And then I was like, "Okay, let us do it. Today we are sitting down, we are making the thing, we are making the first-" And as soon as the first iteration was out, that person was off and running. They just really needed to be supported, to the point where they got to that point. So maybe you just need a cheerleader.

EW (01:03:31):

Or a project that I care enough about to take the time. But I agree with the teaching cheerleading part. There has been so often that all anybody ever needed from me was for me to say, "Yes, that is a great idea." Off they go. And it is like, "You know that my opinion does not matter, right?" But that is all they needed. Yeah, if anybody is in that current state, "Yes. I think that is a brilliant idea, and you should try it out."

KL (01:03:59):

<laugh> Fantastic. I agree. I am sure that you will either have a great product, or you will have a great story.

EW (01:04:06):

Exactly. Kari, do you have any thoughts you would like to leave us with?

KL (01:04:13):

Let us see. I feel like if I were going to leave us with a thought, it would probably be about the power of an invitation. So a lot of times people are like, "Well, how did you end up on some weird career path, where you are making Broadway costumes and spacesuits and puppets and soft robots, and dressing robots in clothes, and making robots out of candy?"

(01:04:41):

And I am like, "It was all a series of things I never planned. Someone would invite me somewhere. They would give me a possibility." And then I would say, "You are right. Maybe I should make puppets. Or, maybe I could be a NASA contractor."

(01:04:57):

And that when we are in a position to invite another person to join us, that is so beautiful. Both in terms of human connection, but also amplifying our capabilities and our creativity, by inviting other people to be a part.

EW (01:05:19):

So I am hearing, "Share your toys."

CW (01:05:21):

<laugh>

KL (01:05:22):

Share your toys. Share your enthusiasm.

EW (01:05:26):

Our guest has been Kari Love, fellow and adjunct professor at NYU-ITP, the Interactive Telecommunications Program. Her book is "Soft Robotics: A DIY introduction to Squishy, Stretchy, and Flexible Robots." And her website is karimakes.com. That is K A R I makes dot com.

CW (01:05:48):

Thanks Kari.

KL (01:05:48):

Thank you so much.

EW (01:05:51):

Thank you to Christopher for producing and co-hosting. Thank you to our Patreon listener Slack group for their questions. And of course, thank you for listening. You can always contact us at show@embedded.fm or hit the contact link on embedded.fm. And now a quote to leave you with, from Hiro in "Big Hero 6."

(01:06:11):

"I fail to see how you fail to see that this is awesome."