 Now I have to say something about a speaker or a speaker is a Matthew Bargati and He's the founder of a company called super-releaser and what they do They are building space suit components robotic space suit components for NASA and They're researching and developing on a field called soft robotics So Matthew will not talk about soft robotics. So I'm not sure but I might give you this hint if you're close To him don't try hugging everything because this robot might crush you with it with its soft delicate hands a warm applause for Matthew I've got to clarify I'm a subcontractor on a NASA project I it NASA doesn't like it when you say I'm working for NASA unless you like are in their offices We just got to clarify that because people who like can laser you from space like so So Einstein was on a train and he turns to the passenger next to him and says does Boston stop at this train? So Why can't programmers tell the difference between Halloween and Christmas? Because 3-1 oct equals 2-5 deck Okay All right, so I'm I'm the founder and lead scientist at that's super-releaser Super-releaser is a soft robotics development company. Essentially. We we solve hard problems with soft robots soft robotics is a new field of robotics is pretty young only about 30 years old and the heart of it is Most problems that get solved with biology most problems that get solved in nature are done with compliant materials squishy stuff and Most problems that are solved in engineering are done with hard Things so hard languages hard actuators things that don't have much give things that don't have much compliance And it's essentially means that half of all potential solutions to problems in robotics have never even been considered And so that's what I'm excited about and I'm excited especially about applying it to human scale problems So I'm gonna be talking about some of what our work is the uses for soft robotics and how to build your arm Okay We're gonna start with a little brain teaser. So Your skull is wired this way you have a hinge in your jaw You have this mandibular hinge and the muscles are connected from your temples down to your jaw if you feel the sides of your head You know clench your teeth you can actually feel those mandibular muscles flexing And so you can actually generate 200 pounds the average person can generate 200 pounds of bite force from their jaw That means if you stick your tongue out right now, you can bite it clean off, which is cool You really can don't try So how would you do this? How would you bite things without a skull? So there are actually creatures out there that do this squids So squids have a beak that's very hard. It's a bird-like beak. It's actually it's really unusually bird-like It's an example of convergent evolution And they don't have a skull. They actually don't have any hard Understructure whatsoever and they can bite things. There are actually many orders of magnitude harder than they are in half They have a body that is about as firm as a tomato and yet they're able to bite through fish and The way that they do this is some through something called functional grading. So up the up the tip of the beak What's called the rostrum end? They have a very hard end the very hard proteins Hydrophobic proteins that don't have much water in them that blend out towards the wing end of the beak that blend out to softer and softer more Water-bearing proteins as the thing goes down into the squids head and that's met by firm muscle tissue that comes on to the wings And then it distributes the force up through the mantle So what ends up happening is you have this very hard tip here Without a fulcrum. It's actually balancing all of those forces of biting across the structure Using distance to balance that force and to get lots of clamping pressure on the end of that beak So what am I talking when I'm talking about applying this stuff to human scale problems? So I trust you can read I'm not gonna Individually talk about these points but just to say the cost of having cerebral palsy is very high and a lot of that is because Physical therapy with one-on-one with the therapist is very expensive and the orthotics the things that you use every day To try and get you more mobility have to be customized tailored to every individual person now that ends up being very expensive because specialists have to do this work so The reason why it's sort of a complex problem. That's not easily solved with a traditional exoskeleton is human beings don't really have a Hard like rigid ball bearing joint and say the elbow. We're not made of these hard linkages They're softer. These are actually two surfaces that are not quite Cylindrical riding over one another. It's not just a fulcrum and Everybody is different every person's a little bit. You know, they've got a little bit of different muscle tone They've got a little bit of you know different something. There are different diameters And so it's hard to make a one-size-fits-all robot in a traditional robotic sense But what's also the case is that? Customizing it for a person takes a particular expert someone who knows that particular robot very well to center its power So if you're trying to have an elbow actuator where there's a pancake motor right here and it's actuating your arm out It actually takes somebody to get that right there as close as they can to the pivot of your elbow So that if it can't come a little bit off of the joint Otherwise it stands the potential of hurting you it can actually do more damage than it's trying to fix So that's one thing that we've been working on at super-releaser is soft exoskeletons for therapeutics So can I get a volunteer? You yes, please come up What's your name Robin round of paws for Robin, please? Yes, so would you mind trying a unlicensed medical device Groovy This and arm could you please put your arm through here? Thank you so much All right, so the problem we're trying to solve in cerebral palsy is the point of your elbow on the point there Yeah Let's move it up a little bit. How's that feel? Groovy so I Wasn't able to take my my compressed air supplies through the airport for obvious reasons But I'll make do with a hand pump So the problem we're trying to solve in cerebral palsy is that you have essentially high muscle tone that you're stuck like this and What we're trying to do is give even pressure to get your arm back out So just tell me what you feel as you go through that So what's happening here is that we have these soft actuators that are distributing force over the arm You feeling a force? Hey Please give her a round of applause Thank you So the concept is one size fits all pneumatic exoskeletons And what I think is cool is that this compliant squishy squid-like way of thinking about mechanics Means that you can do things in these one size fits all ways But what you can also do is sort of compensate for the irregularities of individual human bodies so Other researchers are actually applying this to really interesting problems in like stroke rehabilitation giving people higher muscle tone able to like grip with more force recovering from strokes There's a place called other lab in San Francisco. That's literally making a squishy Ironman suit and What I think is really cool about it is that if you can encourage people to do daily tasks As someone has a physical disability like cerebral palsy in giving them a little bit of extra freedom Can mean huge Lifetime gains because being able to grip crutches being able to dress yourself Enables you to do a whole bunch of things that also motivate you to continue using your muscles to continue refining skills And so little changes to mobility can have a huge change in overall health So Other ways people are applying functional grading two problems are with like Explody soft robots that has a soft structure and a rigid structure So you can get things that are soft and rigid to do particular To solve particular problems and one way this is really applicable is in prosthetics So this is a scan of Hugh hair Hugh Hugh hair's legs So Hugh hair is a double below the knee amputee who works at MIT and he's been working on these projects where he takes scans and He 3d prints functionally graded structures That distribute the force of the impact from the prosthetic onto the residual limb Now the reason why you want to do this is people are made of mostly water People are made of mostly water and any impact does not get dissipated by some cushioning force There's not a cushioning spring inside of the body So when you have a residual limb you have an amputation you have a residual limb your doctor tries to do the best They can to make it susceptible to a prosthetic, but it doesn't mean it's infinitely powerful And so what happens is when you feel a shock say of walking upstairs from the prosthetic socket into the residual limb None of that force is dissipated by the socket to get it registered on to you It also means that waves of force coming up through the prosthetic sort of find all the tender vitals It's not a pleasant experience and what's difficult is that can cause the residual limb over time to break down so scar tissue isn't terribly strong and so it it has this feedback loop a Socket for the residual limb the interface between the prosthetic and you Has to be particularly shaped to you, but if it damages you you change shape And so what they're trying to do is create a system where it dissipates force the socket itself is strong enough to support the force from the prosthetic without Pushing all of that force on to the easily damaged area of the residual limb And so Hugh here is using functionally graded structures to do this 3d prints that have very high levels The white is very high levels of stiffness and the black are very low levels of stiffness so we can choose where the force is distributed so another Organism another mechanical system that does this kind of distributed force to solve problems very well our goat feet So goats don't spend a lot of time looking at their feet They're actually able to put their feet wherever they want say it's slidey rocks They put their feet where they want and they don't stare at them to figure out what they're doing if you ever see an Osama walk upstairs. It spends a lot of time looking down being like okay feet. What are you doing and? It's has to position the foot to land precisely on the step with the right angle or else It'll slide right off and that's because the system doesn't have compliance in it working with actuators and servos It's often difficult to have them embody compliance to be squishy And so it's hard to get them to deal with unknown situations and oftentimes you have to pause a robot in its tracks So it can look and assess its situation before moving on So here are goats also goats and here's a robot so What I'm when I'm trying to solve here is see you notice its feet are hard metal pads That's insane Because it can't possibly compensate for unknown situations if it's moving forward It has to land with its foot at the correct angle if you remember physics vector forces yada yada yada It has to land with its foot at the correct angle to compensate for the force of the forward motion on Whatever it's landing against it's a very difficult problem to solve purely computationally with hard actuators But it becomes simpler you can do it in a stupider way if you just arrange the matter to move in a certain way when there's force and Another organism that solves this very well is bugs so Flies have these things called seite on their feet. They're spatulated hairs What the fly will do is it'll drag its foot down a surface and it could be even be a really smooth surface like glass And it'll drag its foot and the itty-bitty hairs. Those are incredibly tiny That's a tiny fraction of the size of a human hair We'll actually find little cracks and microstructures inside of the surface And it just drags its foot down until the foot catches and it just has this algorithm that is like dragged down Foot n until it catches and then start that process again with n plus plus one And it just does that for all feet and then it does that again flies aren't known for being very smart and So the way we're trying to employ this in soft robotics this represents some of the work I actually got to participate in with a NASA researcher called named Aaron Parnas. He's not called Aaron Parnas He's named Aaron Parnas Where we're trying to do this thing where we can mass manufacture these guys that have a complex behavior that When you put them all together so the individual actuator is really easy to make or the individual Mechanisms easy to make it's easy to multiply them And so you can get interesting behaviors by multiplying that to have hundreds and hundreds of these all working in tandem So the way this works is there's a rigid structure here, you know bringing up the squids again There's a soft structure here, and then it has two mounting points so that you can essentially load the spring on it And what happens is this grips in when force is put pushing down, but as soon as it's unloaded it'll pop back up and The way that these are used is in big blocks What what Aaron's trying to do is he works at JPL to try and make NASA Exploratory robots to climb over rocks and Mars without having to ask NASA all the time. What am I supposed to do with my feet? Because it takes I think it's 14 minutes To relay that signal to earth and then time to get that back and that makes exploring very very time-consuming But and this I didn't flip this gif over. This is literally upside down And so what's happening here is you take all of these hooks and you have this incredibly simple procedure Which is take the paw move it over grip the paw until the paw has caught and then you're done So I'm going to show you how to make your own shot soft robots So these are not sex toys. They are soft robots So I do a lot of 3d printed soft robots. I publish them in the open source I have a lot of stuff on Thingiverse. You can look me up giant. I Thingiverse I am actually writing up a book for make on like every soft robot I can think of so it's all published in one place, you know yada yada by my book So it's not even out yet So It starts with you can actually use 3d printing I figured out how to watertight print with an FDM printer so I can use cheap prints to make things watertight So you can just make molds and what's cool about that is I can make generations of molds that all experiment with different ideas and make Them all at the same time and label them like experiment a b c d e and then cast them all evaluate them against each other and not waste Too much of my time because my time is precious. I can buy robots to do the printing so It's a mold that's sort of like two halves. I'm not tall enough. Anyway, so the mold is two halves and There's a core in the middle and the silicone is squishy enough that once it's cast And so this is a little gasket here that you'll see coming into play in the end when the mold is assembled in the silicone It's cast I can actually pull the core out of the silicone. So there's a structure on the inside of the silicone part there's a structure of a Void so I'm choosing where to put the not silicone and when that inflates it actually distorts the silicone in a certain way and So well happen here is here's the part that that gets mounted on remember There's there's only silicone and air there and then it makes this kind of complex gripper And so I'm going to try to again. Don't have my air supply here. So forgive me Forgive me for not being being macho enough to really manage this but all right gripper I do it do it do it. Yeah Anyway so it's it's able to pick up a big variety of objects and By just being simple and squishy because it squishes around the object And so that compliant conformal nature makes it good it just sort of picking up stuff So the last thing I want to talk to you about is our work at NASA And another way to do soft robots for human scale problems So as you might know space sucks for your body Space isn't very fun to be in it's exciting, but not terribly fun physically So being in zero gravity like sphere eyes is your heart over time The surface tension of water becomes a dominant force and so your eyes sphere eyes There's a particular kind of a stigmatism that happens in space because your eyes Instead of being initially eyes are a little bit egg-shaped. They'll actually turn into spheres and you'll get nearsighted everybody in space wears glasses and So one thing that's really difficult is to make a space suit essentially what you're making is a soft human-shaped space ship and What's difficult about managing that is the softness is there so you can do stuff with your many points of articulation In space to do things like repair the Hubble and stuff But what's difficult is because there has to be air pressure You know you evolved at the bottom of gravity. Well, there's there's a giant air column above your head that reaches out into space That's all of this air pressing down on you And when you let all of that off everything in you essentially swells up your veins dilate and it sucks a lot literally so No, anyway What happens is your body doesn't work well in a vacuum and so what we've done in usual space suit design is to replace that atmospheric pressure That you would otherwise have in a gravity well with pneumatic pressure We have a big airtight suit and we inflate it to like seven pounds and that's that What's difficult is that becomes a giant air spring if you've ever like blown up a rubber glove You'll notice that it's springing so the suit springs back and for doing delicate labor with your hands It actually is really terrible people's fingernails literally do not delaminate from their nail beds and their gloves fill up with blood as They're working on the space shuttle as they used to work on the space shuttle Because the glove design actually causes you to essentially push against this spring constantly Now, how would you how would you solve that like you're currently the gloves sort of like this? So this is like you know, imagine glove is in a vacuum chamber and you're putting your hand inside of the Glove in the vacuum chamber so in the current glove we fill it with air so we just extend the regular atmosphere into the vacuum chamber if you put your bare hand into the vacuum chamber your hand would get all purple and swollen and There are other ways to solve this like you could have a perfectly cast like steel Hand void so there could be a steel block with a hand shaped hole in it You grease your hand up put it in the steel and when the vacuum chamber vacuums your hand wouldn't change shape at all And so you wouldn't experience the vacuum, but that's not very useful for like repairing spaceships Another way to think about it is there's pressure coming out So what if we put pressure coming in you can actually make really? Really small stretchy gloves and have somebody put their hand inside of the really small stretchy glove and then when you turn on the Vacuum the pressure in will equal the pressure out and that's called mechanical counter pressure And so that's what essentially we're doing is Here's the Gemini style glove So that's the big bag of air and here's the mechanical counter pressure glove The mechanical counter pressure glove is made of elastics They're actually biased to be a little bit small and that provides the exact same pressure in as space is pulling out Now the way you get your hand into these is we actually use some soft robotic elements to change the size of the glove So it's a little bigger when you put it on and then you shrink the glove with pneumatic pressure until it's sized to fit your hand and So the part that I'm specifically working on spending most of my time on is this is going to go into space Well if everything works So the idea is not to make a mechanical counter pressure suit in one go But do it incrementally and so we're putting it we're putting it on to the current NASA ISS suit Using an interface from the regular pressurized suit out to the glove that's actually open to vacuum So the the glove protects you from the vacuum pull of space, but it actually doesn't need a bunch of air around your hand So what's interesting is we have this pressurized suit and this depressurized glove and to transition between them What my lab super-releaser is spending the most of their time on is a gasket that takes you from the suit pressure To the zero glove pressure without cutting off any blood to your arm So it's a cool problem and a cool interface and a cool problem to solve And we're doing this with fancy 3d printed molding with magnetic cores and it's What I really like about it is the engineering is complicated, but the materials are so cheap So anyway, I I'm Matthew Borgatti. This is my company super-releaser Please get in touch with me if you want to solve problems hard problems with soft robots and I'd love to take any questions you have Okay, thank you very much Um So I think we will follow the usual Q&A pattern and while you are lining up behind the microphones We will start with the signal angel Hey, so I have two questions from one person and the first one is is there a hand prosthetic version of the Mars robot fly crew like law I'm sorry The Mars the claw explorer. Yeah, is there a hand version like a human hand kind of hand prosthetic Oh, is there a prosthetic that works like that? I've never seen one. I mean, it's totally doable though. Oh I did literally work on a project to do this I worked on a show called prototype this That's how I met Aaron Parnass is we literally made a spider-man suit that used these to a person Had ones on their hands. So I'm changing my story. So We made hands that a person could climb a wall with to give them spider-man powers So there it has existed at one point in the universe And the other question from the person was are shock absorbers for prosthetic legs available or is all the compliance at the attachment to the body So it's a it's kind of a complicated problem. So You might notice the recurve legs if you ever seen the running legs Like the carbon fiber spring legs that's runners in the Paralympics will use they they have integrated shock absorbers It's a compliant shock absorber But the difficult thing is you can imagine to absorb shock You need to move mechanically and so since the force is coming up you need to move down And so if you have just a spring so you can imagine like a piston leg You'd have a spring and every time you step to absorb the shock You'd end up having a very odd gate which might be acceptable might not be but it's not it's a suboptimal gate And so the way it's currently Attempted is by having silicone sleeves that get you between the the prosthetic sleeve which is usually hard carbon fiber and the residual limb is a silicone sleeve But the problem is it doesn't dampen any of the force And so the idea is to not absorb shock by like having the prosthetic go down But distribute the shock to the strongest possible places in the body to distribute it to the strongest places in the residual limb instead of Distributing it over the entire residual limb That I hope for portal like jumping of buildings to Boing another question or the mic? I wish I think we have no questions at the microphone Well, I answered everything then another Yeah, if nobody else So Another question from the IRC is as like designing silicon circuits is quite complicated due to the variability of process parameters How do you predict the elastic behavior of your robots e.g. Like for example by FEM simulation or do you just do trial and error? Yeah, so it It is sort of amenable to finite element analysis FEA But it is actually really processor intensive and you can do things that are like works like models So you can take meshes and you can distort the mesh in it and it's a very nonlinear system Where some distortion changes the wall thickness you can imagine an inflatable thing the walls get thinner And so that changes the dynamic of the system and you have to recalculate What ends up being difficult is having that be accurate to the world is tough And so you can have simulations that give you some predictive power But there isn't anything where it's a good design tool that also features simulation where it's accurate enough to really predict This is going to do that. I did it the the words on the computer match the world And so I I've designed successfully it what ends up happening is you use some of those tools And then I found it's best to have frameworks where you can iterate where it's simple to modify Because the the simulation will never be as informative as experimentation There are technical documentations the amount of stretchiness in the silicone and you can predict this to a degree But I found the best way is to iteratively prototype and lower the cost of making a new prototype by making it really quick and really cheap and that Ups the amount of prototyping you can do Thanks, and as a follow-up like how much does the material properties vary like the variation really high or So silicone is very consistent So I use our TV silicones room temperature vulcanizing you can get them their two-part silicones You mix them together and the the chemistry on them has advanced to a point at which you just 50-50 mix them and they cure And their properties are very They're very consistent Their variability is documented it's usually fairly low A Some of it requires a little bit of process control where if you're very diligent about it You need to make sure that there are no bubbles in the casting and that your casting is dimensionally accurate They're all these mold making tricks and techniques They're sort of like Adam Savage model making tricks and techniques of like you got to tip it to the side to make sure the Bubbles get out in this direction and stuff, but it's very consistent Okay, then we have actual people we had microphones and not that the people on the internet aren't actual people They're not right, but but these people are around here and Since we have a lot of time you have left where we can have more questions. We can have more questions from the internet, but first Oh a few questions from the microphones. Yes Mention before and that the resources are really cheap. Where do you get your resources? I don't know in comparison to what I don't have any Funds funds to research with where do I get my materials? How do you deal with getting a specific shore and do you work with shore? I don't know what your material specifications are and have you mixed different? Components together. Well, how would you advise we do this at home? Okay, so to break down Where do I get my materials? How much do they cost? Do I combine silicones and sort of what shore what like range of durometers am I looking at sure? I'm sorry, I have done a lot of research and I haven't found any answers. So I'm asking you Conductive silicones. Have you found any do you have any solutions? So Conductive silicones there if you look at dielectric elastomers, and it's spelled D I E like die electric elastomers Dielectric elastomers there is some documentation on them They there tend to only be conductive in very high voltage ranges like plus minus 500 and up but They do exist I I tend to discount them because it's sort of not my it's not my field There are some people who are working in software robotics who use shape changing elastomers for that that use an electric shape change effect The materials I'm using I you have an FDM printer. I use an ultimaker to I use whatever cheap PLA I can find on eBay I Use Sainsmart a lot the silicone that I use is smooth on I will use smooth on dragon skin 10 and Ecoflex double o50 They are about a hundred dollars a gallon But it is much it they are much cheaper than the technical silicones you'll find for medical research And medical implants those end up being a hundred dollars a pint the the adhesives I use there's a Adhesive sil poxy that smooth on makes you can also use epoxy silicone cocks that you find from Home Depot and stuff like that the shores that I'm using range from double o50 to Shore a 40 I'm usually around the shore a 10 region you can thin silicones out with silicone oil and you can also buy thinning agents from silicone manufacturers and Ecoflex is miscible with dragon skin because they're both platinum cure silicones that use the same Filling agent and so I mix them oftentimes at a 50-50 ratio to titrate the durometer Smooth skin doesn't have any European distributors. Do you have any solutions? Oh Amazon import tour for physical objects, I Don't have a solution for for exporting. Um, I Would call them up and ask nicely Thanks All right, so we saw you using that hand pump to close that yes hands claw thing Yeah, so I'm just imagining that on a on a human being for example, how much air pressure what you need to actually make it responsive hand prosthetic and Like how much air pressure would you need for tasks like lifting a jug of milk like it like a gallon of milk For example, and would it be feasible to install all the hardware? You would need to generate that pressure on a human being on earth and yeah post a space relevant questions So my devices tend to articulate from about 10 to 20 psi I use compressed carbon dioxide canisters which act have about the power density of a lithium polymer battery So if you look at grams per joule, they have about the same power density So I tend to use them especially because the torque is instantaneous They come if you've ever reinflated a bike tire, so it's those emergency bike tire reinflators Or if you're San Franciscans and like doing whippets, it's those things and I I Think the systems that I've made to compress them into like human scale usable things You can't take compressed gases on planes whatever But I think that the the power density is appropriate for articulating objects in this domain People have worked in them other lab has worked in that Getting lots and lots of force out of things like if you want to pick up a jug of milk with just Compliant mechanisms like you're starring only silicone That's a more difficult problem because you have to stiffen the silicone to the point at which it can like Resist the weight of the gallon of milk But a way to get around that problem is to put spines inside of things So the way nature has done that is when things need to have a lot of torque they'll actually Put spines in things and the spines will help keep distortion like linear distortion from being a Detriment from the total amount of force that you have in the system and so they are you know snakes do that snakes solve that problem with spines, but You sort of have to choose your battles. It's not a perfect solution for all problems It's a good solution for compliance interfaces But you have to sort of choose your battles and I look to nature to do this to figure out what mechanisms are appropriate Thank you. Okay. I heard we have more questions from the internet from nice people from the internet So if they could all from a fridge, you never know We have one question here How do universal Jamming grippers compared to articulated grippers. This is quite special. I guess yeah Okay, so universal jamming gripper is essentially balloon filled with coffee grounds If you ever bought vacuum-packed coffee It's like a hard brick until you pierce it and then it turns all soft And that's because without any air to have the individual grains the granules to move past each other the thing solidifies And so you can get conditional stiffening by pulling out or putting in air And so universal jamming grippers are rubber balls that have coffee grounds inside of them And then when you suck the air out they stay in whatever shape they are And so if you squish it over an object then suck the air out It'll pick up that object because it's stuck around it And so it can even be a straight-sided object like a dice, you know a dye dice Name for one of those things So even if it's a straight-sided thing being rubber It does there isn't any draft on the individual Random number generator to actually pick up or it can't drop right out because of the friction on the walls So the advantages of those is they can pick up a huge variety of objects even things that are like really flat on tables The disadvantage is they get pierced pretty easily And that's a problem sort of all pneumatic compliant systems But they get pierced pretty easily and there are categories of objects that they can't pick up So anything that has like a handle that you have to get all the way around Because there are disadvantages for using the jamming gripper too So if it's really tall and you want to squish it over a handle you need enough space to squish it over a handle It'll actually distort a lot because the more space more distortion And so as you pick it up, it'll actually be like And let go of heavier objects one of the advantages of like the software botics incorporated gripper is that the the Octopus tentacle style grippers the digital Hand grippers are good at like wrapping around things like a couple of times or wrapping around things with handles and picking them up You know their their disadvantages is they're not very good at like picking up a penny. They're very they're very bad at that You have more questions for us. Yes, I have a question as we have still quite some time I guess it will take a bit So how does one build small lightweight sources of banana prunett pneumatic or hydraulic pressure Got it. Yes, really small Yeah, so that's a problem. I'm actually trying to work So you there are ways of getting small amount of pneumatic pressure with piezos So what you can do is you can have a closed volume and goodbye computer anyway So you can use piezos that have essentially have two check valves and the piezo is pushing into this airtight container and it'll push a little pressure out of the check valve out of check valve One and pull a little pressure from check valve two and they can articulate very fast and it'll generate Pneumatic pressure the disadvantage is it's usually not that much and you have essentially a spring system in there Where if the pressure the back pressure is more than the pneumatic that sorry than the piezo can overcome Then you get it peeks out at pressure And so that's a way of getting really really tiny pneumatic volume or pneumatic stuff going on It's essentially you imagine taking a piezo speaker and then putting a couple of check valves in a thing You can have it like a printed enclosure for it And that would be a really tiny way the other ways that it's usually done is you have a diaphragm valve diaphragm pumps the small cheap ones that you can get from China are these diaphragm pumps that just use a rubber diaphragm and they push pistons that are actually all on a pivot and So they'll go around and the diaphragm will push in and out and check valves will cause the pressure to continuously go out And so that's the way that's done Hydraulics small hydraulics are not super my field I don't know that terribly well my solution for that has actually been to use pneumatics To displace hydraulics to sort of like you could imagine bubbling air up into a soda bottle And then having a straw with the hydraulic end going down But that's how I handle hydraulics Okay, I have just realized how long you've been standing there. So the microphone, please and why didn't get anyone human chair or something? If you let robots roam a forest then they break something after like five minutes and they will destroy themselves pretty soon So could soft robotics also be a solution to this problem? I think that it is so the the you know to like wrap it up like To wrap up that that question in sort of a bow Boston Dynamics makes this really expensive Robot called big dog and I'm I'm not here to like Boston Dynamics Because like who am I to talk I make I make these like I'm not super sophisticated But one of the disadvantages is it's a it's a complicated robot with lots of things to go wrong and One thing that can be very difficult and having it operate for an extended length of time is that it has to make all of its guesses exactly correctly to not like topple over and it's supposed to like keep soldiers from dying and going into dangerous situations instead of a soldier but There are I have had reports from people anonymous sources That people have literally died trying to rescue the robot because it was on their heads if the million dollar robot exploded and Yeah, so the solution I think that's appropriate for it is to choose your battles where you put Actuation where I think there can be a lot of passive systems that actuate that like have a little bit of a bias Like your foot is an example of this where your foot is doing a lot of passive compensating for like what the Dynamics of the like the walking downstairs or walking across rubble Your foot is compensating for a lot of that just by being soft and you bias it in small waves with the tendons and muscles that are in it And so that that's my belief is that you choose where you've got springiness and compliance And then you match that to where you've got your power for like locomoting and for you know hauling stuff Is that sufficiently answered does that feel pretty good? Thank you Okay, oh we have another internet. Are you a signal angel? Because we can't answer questions that are not asked on the microphone because we need to have them on tape But we have the signal angel and she has another one for us. I have another follow-up question Yeah, we still have some time So if you would need to trade off panoramatics Versus Hydraulics, I'm not possible. I'm not an expert hydraulics. Yeah. Yes. So which would you use or what was the use case for each of those got it so Hydraulics are really good for things like mining So you have like a drill that's being turned by a hydraulic system And it has instantaneous torque and you can get a whole lot of force and there are relatively low line losses because of friction that That's a really good use case for hydraulics when you need instant torque and you don't want springiness you can design springiness into hydraulics, but since water is incompressible a Piston that's being driven like a hydraulic actuator is exactly where you put it So you actually have to like to compress water water is technically compressible But it's like on the order of like point 101% but you actually have to sort of to squeeze the hydraulic cylinder and get any spring out of it You actually have to like it ends up distorting the metal around it or blowing the gaskets so that Works interestingly when you talk about compliant systems because you usually don't need as much like this isn't going anywhere when dealing with Compliant systems, but there are situations in which you want really instantaneous torque and things like that for like You know being able to really quickly Change the pressure inside of a system where you're like someone has spastic motion in your their arms You're trying to exactly deaden it by returning pressure that exactly equals the opposite of that waveform And so it's an interesting use case, but it's difficult to implement I Use pneumatics partially because I can be like wasteful with it where I can have things that are leaky And it's just like I'm I just have a monitor on how much pressure is in the system And I don't have to worry about how much I'm putting in as long as it loops back to here's how much system pressure Here's the state of the system It doesn't matter where the dial goes as long as it stays at that state It's fine, and so it's easy for me to have slightly sloppier systems that I can sort of Fudge around with and create quick prototypes on and don't have to hunt down leaks And that's why I use pneumatics because that that is a nightmare in hydraulics like hunting down leaks and stuff And I don't want to be covered in hydraulic fluid every day Okay, I'm took first you in the back. I think because you have been raising your hand for so long Okay, you said and your your company is a NASA subcontractor And I think you said that you were using or that those those gloves that you made will be used on ISS and space suits if if everything works if everything works, so Obviously you believe in your product and yes sold it very believable to me But my question is and does NASA expect this to be the future of space suits or are they just paying you to test around? Noodle about that'd be great. That'd be great job What'd you do I messed about Did it work? No, that's fine So they're they're investing in this because they want to solve problems in space suits and they think mechanical counter pressure is one of the ways to do it Compliant materials like play techs made the first functioning space suit that ever went into space, you know Not Northrop Grumman not bowing play techs international latex incorporated and it was professional bra seamstresses that designed and built the Gemini space suit so The science like the down and great nitty-gritty science on soft materials has always been Difficult to evoke in physical stuff But like having a gist of it understanding the materials from working with them and applying to them to practical problems has sort of been the thrust of most space suit technology and so NASA is like well you guys you work with this soft stuff Can we do a new thing that hasn't been done before with space suits and If it works it could improve a lot of things and solve a lot of problems But what's difficult in in brand-new systems is you're like looking at your ideals and then you find out how it fails like the externalities of it and so we're exploring a lot of interesting things to try and solve these top-level problems and the designs are Evoking these sort of mid-level problems and we're sort of hacking towards idealizing all of it But the performance we see on the gloves like the dexterity in vacuum chamber tests and the amount of pressure that you see like the Even pressure across the hand without like causing fingernail delamination is really promising right now And so it's encouraging NASA and it's encouraging us to keep pushing with it Yeah, you you on the microphone We try to build some soft rob robotics in our maker space in Milan and it's very difficult So I think that without having a design model to actually design the things yeah before to Build it. It's very hard. Yeah So what do you think about the design process of soft robotics? Yeah on Unfortunately, there's no there's no way I can give you like a data dump It's it's almost like if if you had to still copy see programming out of books Like if if it was very hard to like if the internet didn't exist and you still had to copy see off of books You'd have thousands and thousands of people being like why doesn't my program work? I've read the entire book and it's like well You had a trailing slash here or something like there isn't a good guide to give you an intuitive knowledge of it without Playing with it a bunch because there's no real good well-spring of people who are jumping in to answer your questions Like there's not a soft robotics forum. Well, there is on my website But I actually have if you go to superreleaser.com slash forum. I Actually do have a software box and I will answer your questions but The difficulty is there isn't a level of knowledge up there to solve sort of meta problems of like What's a good way to think about this? What are successful models of soft robots and so like getting a base understanding to know why this doesn't work and know What a successful software robot tends to be designed like and good design processes are thin on the ground And I think that makes it a really huge ramp. You have to really get up the ramp Of the learning curve really quick to be able to produce results to keep going And I'm trying to solve that by publishing open source robots, you know Make working designs and be honest with about the design process and then publish, you know What works and here's what to look out for But before that's thicker on the ground and before people are available to just answer questions from people getting into it There's almost no way you can do a data dump to somebody be like just just read this book And then you will know exactly what's wrong with it because it Like many design problems. It takes having a detailed knowledge of the system in place Okay, I think we have one or two more questions from the signal angel again I use these working well, and we have still a lot of time left. So All your questions Yeah, people have been really fun and they also want to thank you for the talk as it was really cool And yeah one sees that you really have fun with these research So the one thing people are still interested in Can you expand a bit on the one million dollar robot being saved by the soldiers? Because that was really quick ah Sort of so I Don't want to I don't want to talk too far out of my depth because you know, I'm not in the military I have friends, but I don't want to just be like here's how it is everybody so In some ways and this applies to all the equipment that you use if you're like an Electrician like and you're servicing something you're responsible for it. Even if someone Directly working with you might say well that that exploded, but it definitely wasn't your fault up the chain it's your fault and So when you introduce expensive equipment that people are nominally responsible for if it fails You have to ask you have to find some way to justify for them Why they're allowed to let it go and so when some when there are some systems that in star incredibly complicated and incredibly expensive equipment Like a robot that's supposed to do certain kinds of field work If it gets lost on your watch You are nominally responsible and therefore have to take some of the heat for it And so I think in certain situations when robots aren't a hundred percent reliable It ends up with people having to take responsibility for like saving the robot Which is antithetical to their initial purpose and so my theory is if you can get robots less Expensive if you can change the manufacturing principles of robotics Which is sort of one angle that I'm trying to go with with soft robotics is that it lowers that Expense value so raises that like disposable robot. Just it can blow up. That's okay. We can get another value changes sort of that equation But they are not like why liable for them or like have to pay if it breaks I don't think soldiers are individually reliable Responsible for I don't honestly. I'm the last person you should ask about how the the billing system works in the United States military Okay, thank you the words Okay, oh we have more questions. Yeah, you start popping up at the last seconds of the quest Question and answering from the signal angels every time. I think you might have found out how not to stand all the time The microphones again, yeah, I have a question about Going into production. Do you have any example of a project that reached mass production? mass market production of a soft robotics project so The best example is actually nerf So Nerf uses a ton of soft compliant mechanisms to generate pneumatic pressure and to say choose Which dart gets shot in in guns? They actually use a lot of soft compliant systems where it's like, okay So you squeeze this and it conditionally will inflate that and that will go out to this particular rocket Yeah, nerf gun and and compliance systems are everywhere Like sales are compliant system. They conditionally stiffen based off of conditions So like based off of the wind and their direction They will stiffen because the the fibers have high tensile strength and you can actually choose the shape of the sail Based off of how you trim it and that changes that changes how much of a wing It is and various other properties that allow it to go faster or slower and to be able to like Finally tune like as you go over waves. That's one of the points of having a Nerd talk anyway, it's one of the points of having a Line on the edge of a Viking sail So if you ever see illustrations of Viking ships, there is a line going out to the edge of a square sail And you might be asking yourself why the hell would you do that? Why wouldn't you just let the square sail be the square sail? And it's actually because the way Viking ships are designed. They are not Self-riding they are actually very tippy and so when you go over waves the boat flexes a little bit and You need to change the shape of the sail and what direction it is taking the wind as it goes over waves by Changing how aerodynamic the sail is or else you flip over and the reason I know this is because there was a book Published from an anthropologist who studied Viking craft made an exact copy of one and kept flipping himself over because he Forgot to have somebody trimming the sail Okay One more question on the one more all right one more. I don't know we have still Five minutes left. So if you are if there is he has more questions of course first of all him Hello, so you have soft actuators. Do you also have soft sensors? Oh Soft sensors so that that's a thing that I'm working on as a background project So there are lots of ways to use micro fluidics as sensors to have like a thing filled with a hydraulic fluid In a tiny little channel and a piece of silicone and read the pressure out the other end, you know with a little You know a micro sensing device One of the difficult things about soft sensors is it's difficult to Compensate for the flex of the system when calculating how much forces on it because you can picture if you made a flat Sensor array and you wanted to detect how much pressure a person say you have a person was with a amputation Residual limb is inside of a socket you're trying to measure where the pressure is you have a flat sheet of sensors And you stuff that into the socket and the person stands on it the sensor Array is actually going to be detecting a lot of forces because of the particular order in which it bent Not directly because of the amount of pressure that it's experiencing so one thing that is really difficult in sensor design is coming up with things that the orientation like the Way the sensor is accidentally stuck to the side of the Cup in which you're trying to measure the force or whatever that is It's difficult to compensate for that to make sure you're getting something accurate And so one of the ways that people do that is by embedding stretch sensors in soft robots And so they have like a linear measurement of how much stretch has been done With the molding that I do this is sort of an example of it with the molding that I do I can choose where to put individual electronic components inside of a soft thing and so this is a ring that has a Trinket pro inside of it and a couple of RGB LEDs and those are In a specific spot in the silicone because they have a little rider board that actually slots into the silicone and locks in Mechanically and so you can you can if you choose your battles where with where you put your sensors You can have them integrate with soft systems, but it it ends up being a confounding problem But one that a lot of people working in prosthetics and in spacesuits are working on and doing cool stuff it Okay, so the question is signal angel. All right. We have more for us. Yeah, sure people Time to get creative So the last question from the IRC is Do you and can you build pneumatic computers like state machines? With this yeah with the stuff or do you always have to use a computer like? Interface. Yeah, that's actually a question. I'm really interested in because the idea of doing pneumatic state machines of having like something biased where you can have a series actuator where it's like this thing will only inflate under these conditions and so you have like This is the way octopus arms work octopus is really cool. You end up taking a lot of soft robotic stuff from sea creatures So when octopuses are panicked and trying to move all of their legs at once They actually don't have enough processing to really do all legs flowing at the same time And so if you see an octopus like pick up and run it will actually limit the zone of solutions It's trying to find by creating straight sections in its legs And so it'll actually create joints so it can physically run and the the way it's just it has a distributed Neuronal system so octopus individual limbs actually have their own processing centers and so you'll notice there they have a distributed brain and So one thing that's really cool is that you can actually look at with electrodes You can look at what is happening in an octopuses limb as it's gripping something and it's actually that you'll have these relay centers where It grips something with the end of the limb and it'll be controlled by this functional relay center And there'll be a signal sort of coming down and coming back from like the their arm will have a relay center like the brain in a brontosaurus butt and There will be like little nodes up and what will happen is the nearest node will start the flex and If another node gets hit like as it's wrapping something up if the next node gets hit It'll continue wrapping, but if the next node doesn't get hit It's gripping a small thing and it can discontinue the wrapping motion And so what it's doing is it's sort of sending a signal back and forth and asking where do I need to stop the wrapping? wicka wicka so To try and simplify the problem space And so I think that would be really cool to do in soft robotics to have these passive systems Where you don't need a lot of signaling going on to just have PASIC pneumatic like That there's pneumatic logic inside of it I think I've created a solution for that that involves like this two-part molding The difficult thing to do with pneumatic logic in soft stuff is you have to create check valves and Check valves are really hard to create without a spring and when you cast something you don't cast it in a spring orientation It's hard to preload and so it's very difficult like if you imagine a heart valve You know the heart valve opens up. It's a it's essentially a heart valve that looks like shy loot it's got like three actuating pieces and They're actually all bigger than the whole and so they fall back towards each other to create a checked valve where pressure can't come back on the heart It's very difficult to do that in casting because they open up like you cast them in the opening orientation There's nothing that springs back to cause them to close But I think that I have created a solution for that with some complicated two-part molding to create things that Get biased by another part that gets put in but that's sort of like interesting theoretical soft robotics that currently doesn't pay the bills so Okay, again, thank you very much for your interest We are