 I'm going to talk about robots, but a particular subclass of robots called soft robots. And the question you can ask is, what are they? And I'll tell you something about that. The more important question is, who cares what they are? Something about robotics, generally. And there are several people in the audience who know more about this than I do because I'm a professor that doesn't deter me at all from talking about the subject. And then I'm not going to talk about that at all. So let me start, generally, with the subject of robotics. A little background of what is that. Those are gold coins. There was once a time when nations competed for wealth in the form of natural resources, gold, land, whatever it might be. And the world has changed. And now what nations compete for is basically skilled people, information, technology, and knowledge. And that's terrific. So there's a competition. And what's the competition really about? And I think one can argue that when you think about the competition between US, Europe, and China or US, Europe, and India, what we're really competing for is none of the above, but we're competing for jobs. And more than jobs, we're competing for good jobs. And what does that have to do with anything? Well, this is often phrased in terms of high wage versus low wage kinds of jobs. That's that's gone because the wage differences are not much different now. Skilled versus unskilled, where you find skilled people everywhere, but this is the part that I think has to be interesting. And that is, what about a world in which it's not people competing with people, but people competing with machines for jobs? And that's robotics. And that problem can be phrased in one of two ways. It can be either people versus machines or people with machines. That is, machines that augment the capability of people. And the new element in this, I would argue, and many people would argue, is robots. And there are lots of folks who will say that if you ask, what's the next big thing after the web? The next big thing after the web, arguably, is machines that begin to rival people in their competence. So these are just pictures of certain kinds of machines, robots you're familiar with. This is an assembly line. This is a drone. This is the Roomba, which is the world's least successful method of cleaning floors. This is a surgical robot, which is one of the world's most successful methods of increasing billing rates for procedures. This is Mars Rover. And I'm going to show you some things here, which come from the military, who has done a lot of this. Now, there's a word you need to know. When you go to cocktail parties, it's important to know certain words, because you can use them and you appear to be an expert, even if you know nothing about the subject. And one of them is cooperative and non-cooperative, as in robotics. And a cooperative robot is one that works with people safely, and a non-cooperative robot does not. Let me show you an example of a non-cooperative robot from, this is from Boston Dynamics, which is a company that does this kind of thing very well. The sound needs to be up, because it's disappeared. There we go. So this is the tether. That's a concrete block. And this is not what you want as a babysitter. And I want you to look at the next, I think it's the next one that's really the thing that's interesting. Look at the beautiful way in which this balances. Shifts its weight as this center of mass moves. It controls itself. And the thing that you need to understand is it does all of that without a brain. This is just gyros and accelerometers and local feedback loops and things of this sort. I can't tell that that's not alive. Now, it's tethered, and there are things like that. But its balance is so terrific. So the interesting question now is you take mechanical systems that are rapidly beginning to equal animals in certain aspects of their performance. This is the biomimetic part that we're talking about. And add to it artificial intelligence. And I see no reason why we shouldn't have things that are the functional equivalent in 10 years of a cat. Now, a cat is a particularly stupid animal. So this is not a great claim to something or other. But nonetheless, it's really important. All right, so now what are soft robotics? Soft robotics, if you looked at that thing through the concrete block, it's all metal pieces. You don't want to get in the way of it. It really doesn't know that you're there. So soft robots are robots that are designed in such a fashion that they move. They do stuff mechanically, but they do it in a way which enables them to work cooperatively with people. And this program started here. And you say, we've gone off track. How can soft robots start with a coke can? The answer is the problem that we were posed, in this case, or the problem we posed to ourselves with DARPA, was to make something that started looking like a coke can and unfolded and could walk across the floor, find a crack under the door, crawl under it, and reconstitute itself on the other side so it didn't look like a coke can. This is not a specific problem that I happened to be interested in, but you can imagine why somebody else might be. And what is our inspiration for this? And this is where soft comes in. This is a soft animal to act as our source of inspiration. One of my favorite creatures, it is a octopus. I particularly like to eat them. And the reason for that is that I have an ideological belief that one should only eat things that are more intelligent than you are. And this qualifies. But look at the way it's pulling itself through this very small hole. I can't do that, and I'm going all the way to the end because there's a little bit at the end that I particularly like. I don't actually know what portion of its anatomy that is, but the brain is plastic, and it's somewhere up in there. But look at the look of intense self-satisfaction on this. So this is where we started. And those of you who are in the robotics business will recognize there's something interesting here. And this is an uncooked egg. This is modeled very loosely on a starfish. And the problem of picking up an egg is not a trivial problem, because if you're doing it with hard grippers, you have to have sensors. You have to have feedback loops. You have to do a series of things of this sort. What we have here is simply a inflatable structure that I'll show you in a moment. That's the suspension element. And here's a single pneumatic input line. And so when this structure comes down and grips the egg, it cannot apply pressure at any point on the egg that is larger than the pressure that you use as an input, hence as self-adapting, hence the properties of materials have taken over from the usual control feedback systems that you find in most robots. How has it made simplicity itself? One has a structure which has the characteristic that there is a flexible but non-extensible element here and then inflatable element here. As you blow up this, the system has to bend. And that's what's going on here. And I'll show you some complexities of this, but that's the basic idea. Now, there are some things that I just want to sketch for you. Here is a picture of the structure. It's a series of little bladders that are connected by a network of channels. But if you look at this system flexing, you notice something very interesting, if you look carefully, which is it starts from the tip. And that element of starting from the tip could be done in more complex mechanical systems, but would require a lot of control to do. Here it occurs naturally because this is an example of a non-linearity in a material system called, in this case, snap-through. And there are many ways in which you can build properties of materials in such a fashion into these systems that they replace control systems. How's it done? Simplicity itself, you build a 3D mold using a 3D printer. You mold an elastomeric plastic in this, and then you simply glue stuff together. And I'll show you how it's done in earnest a little bit later. But the polymer that you use here is the same one that's used for facial implants and breast replacements and things of that sort. So it is quite soft and quite elastic. Now, you can also get fairly rapid response. And these are systems that, in this case, are still a long way from the apostionata, but they're moving in that direction. But they have about 100 millisecond response times. And I'll show you an application of that in a bit. And then you can move to systems of this kind, which is the first stage in the Coke Can story. This is a system that has one, two, three, four, five bladders. Here is the pneumatic tube that underlies it. Here is the bottom of our hypothetical door. It's a glass frame here. And there's a certain poignancy in this whole video. People, when they see these things, say, oh, that's really neat. It's almost alive, but it's really creepy. And so what it's going to do is work its way under here. And a point of interest in this is that it has, in this case, show two gates. One is one in which it's erect, which we call walking. And this particular gate, which we call uching, because that's a highly technical term. And it will make its way under this. And then it will stand erect again and do whatever it's doing. There's no reason to go through the entire business, but you get the idea. Now, let me show you a couple of things that are interesting about this, this class of things. And I'm just really sketching. This is something that you've seen before. But this is a model for a hand. By modifications to the structure, it's possible to build in features that look a little bit like knuckles. And one of the neat things you can do here is to ask questions about biomimetic systems. Are you better off having three fingers or four fingers or seven fingers or two thumbs and two little fingers? You can make all of those structures very easily and test their ability to do things. And that's quite interesting. But there is another thing that you can do which illustrates one reason why these are attractive as a complement to what's been done in classical robotics. That's a hammer. This is our little robotic hand. And so we whack it. And at least my response to this is always to say ouch. And to show you that this is real, it's real. But if you did that with a small hard robot, you'd smash it. These things almost can't be broken. You think they can be cut, which hard robots can't. But you can throw them off a building and it turns out their terminal velocity is not enough to break them on impact. So what can you do with that kind of thing? And here's an example. This is basically a Sears and Roboc glove, a department store glove, with these fingers that I showed you playing the piano put on the outside. And why does anyone care about that? And the answer is that this is an example of one of the large areas of application of this which is in surgery and rehabilitation. The particular problem here is that if you have a hand and you have stroke then it's really important to manipulate the hand otherwise you develop spasticity. And that's presently done. The manipulation is done with a nurse who is expensive and gets bored at the process. So if you have this kind of thing, what you do is you simply, there's my cursor. This will do any sort of manipulation you want in a way that's very compatible with human beings and also very inexpensive. And so there's an opportunity to do all sorts of things in which these actuator systems work cooperatively with people. And these are simply examples of some applications. So in addition to flexing, you can use this sort of system in many variants on it to increase the strength of the grip or to give you greater dexterity or to do things of that sort, which you can't do if for one or another reason you have a neurological problem or a stroke or something of that kind. And these are really proving to be very interesting and useful. Now, just one small issue in this subject. And I want to return just so that you can look at this. I'm not going to spend time on it because it gets to be sort of technical. But if you ask what's happening here, what happens and what causes these to curl from the end is that the first thing to blow up at the end, the very tip there, is the chamber that has nothing on one side. So that blows up. When that blows up mechanically, the next easiest thing to blow up is the one that's next to it. You've all seen this phenomenon when you take a balloon and you blow it. It's very hard to blow at the very beginning, but then it forms a bulge somewhere. And as soon as it forms a bulge, that bulge propagates. It's exactly the same phenomenon. But one of the real attractions of this kind of biomimetic structure is that it gives you the opportunity to take advantage of the nonlinearities in materials to do things that you really couldn't do otherwise. And here is an example of one of our current models. This is, as you can tell, a spider. Spiders are pretty interesting creatures for all sorts of reasons, but one of them is that much of their actuation comes from the way in which they handle these joints and the mechanics of those joints. And I'm not going to show you the details, but again, but it's another simple pneumatic system in that case run by a fluidic system. And one can easily model, not easily, but one can model those kinds of systems. And what you see there is a thing which has six legs like an ant, but it uses the same joint actuation and movement system that a spider does. And that particular one is made out of sodas straws, the kind of thing that children use with some other stuff. But you can also do more elaborate things, and here is water strider. And this is actually pretty much the way a water strider works. It supports itself on four legs and then it has two legs that scull. It has to have hydrophobic patches so that it stays on the surface and it has to have characteristics that enable it to not sink. And this is lightweight, can be made hydrophobic and you can manipulate those arms as it does, more or less. Now I want to show you briefly one last set of videos. And the idea here, at least I hope, let's see if I can get these two plodding, here we are. The idea here is that one of the things, particularly if one is talking about the developing world, one of the things one would like to do is to find science that leads to technology that leads to jobs, goods and services relatively rapidly. And what's been so interesting to me about this is that our first paper in this was published in 2011. In 2013, we started a company which is called Soft Robots. I have had nothing to do particularly with that, but they have gone off and looked at grippers. And I want to show you what good industrial engineers can do with this kind of thing. One of the questions that came up was, can you lift a weight? And there is this kind of gripper lifting weight. Potato chips are actually hard to pick up because you tend to pick them up in a way with hard grippers that they either have to have elaborate controls or you have to do something that enables you to avoid crushing. Picking fruit is an interesting thing that requires delicacy, picking up this cup and then the task we all face every morning of picking up a cactus. But I'll show you just one other brief section of this. Oh, this is very important and those of you will recognize it. This is dexterity and there's a neat example of dexterity right here where one has a ping-pong ball and you reach around and you manipulate the ping-pong ball. Notice this is on the end of a conventional robotic arm but it catches the ping-pong ball quite successfully and I think this video, you know, this is something slightly different but that's something else which is pretty hard to do. Here, this is interesting. For these end effectors, what one is always going to have probably is a hard arm and you have to change the fingers and you just saw that system change from three fingers to four fingers. It is very easy to do so that what one will have is an arm and then a tray of different kinds of effectors and the system will go from one to another. Now that gives you a sense for where things did this. One last slide or maybe two, one last slide here. Who did it? And a lot of what one thinks about what one thinks about new areas is do you have to have a synchrotron or do you have to have a light source or do you have to have something like that? And I hope you've gotten the sense that the work that's involved here actually involves not much more than very simple ideas in mechanics plus simple polymer chemistry and molding. But what interested me was that if I look at the group of people who've migrated to this, a couple from the US but only a couple and then Spanish, Singaporeese, Italian, Chinese, German, Canadian, several Iranian students, they're very good. Several Turkish students, very good, Chinese, Canadian. It turns out that everybody is interested in this and almost everyone is able to do it. So it is a technology that is both scientifically interesting but also technologically very useful in which the barrier to entry is very low and that I think is relevant to this particular meeting. So then let me use one last slide to summarize. I said that I would tell you what soft robotics are. I have, they're soft, they're robots. It's not too difficult. What are they going to be used for? And the answer to that is I think that at least initially they're going to be used for tasks that hard robots cannot do. And the wonderful thing about a biomimetic study is that if you think about all the organisms out there, insects and single cell organisms and dual flagella, algae and animals and birds and whatever, they're a very, very large number of biologically successful architectures to model. I emphasize that what we've done here has nothing to do with the mechanism by which biological organisms work. It has to do only with mimicking the function or motion and that turns out to be a much easier thing to do. But what it opens is the door for robotic systems that have the characteristic that they can operate safely and cooperatively in contact with people which has been a barrier which has been a little bit difficult to get over. So collaboration is very important. Another very important thing is cost because these things basically don't cost anything. That again is something which is available to all levels of economy. And it has the interesting characteristic that for some of the planned applications as in search and rescue think Fukushima. You would like to have something that could crawl over rubble in water in a high radiation environment, see what's in there. And by the way, when you're finished you just cut the cord and you leave it there because you don't care. Same thing in a burning building. And you don't want to do this with a 50 to $100,000 hard robot but you would have no trouble abandoning one of these little rubbery things. And then applications, we are a group that is very interested in applications because I think they focus the mind and the students like them a lot. So what are we going to be doing? Here are some surgery I've mentioned, surgery and rehab and prostheses, obvious for cooperative systems. Agriculture and the issue here is to take humans away from jobs that are really hard physically and tedious. Search and rescue I've mentioned as in single use, hazardous tasks. But interestingly the one that turns out to be the really big deal is gripping. And you say who cares? And the answer is Amazon. Because if you look at these big end-of-chain consumer product stores you write in or you type into your computer the idea that you would like to have a can of lube oil and some potato chips and whatever you happen to be interested in. And it turns out to be very difficult to find a non-human system that will pick all these things up and put them in the box. These have the characteristic that they can do whatever you want. And I remind you that if you go back to that, the picture that I showed you, I want to make this last point. You go back to, let me, I'll find it again. You go back to the example which was this, which is what the industrial engineers came up with with their solutions. And the point I'd like to make about this is that what you see there started with the idea of a starfish but actually there's no biological system that looks like that. So one of the interests and charms in this to me is that you end up working from something that is a natural structure that has been successful in a Darwinian sense but you may well arrive at solutions to problems that are I think only vaguely based on those biological issues. That is the final answer may be quite a different answer. And so in this particular case, what you see is something that has fingers but so far as I know there is no biological system that operates in this fashion. So it's a very entertaining area is one that spans from observation of nature to production of technology. It is, it turns out really easy to do and there are now a gazillion people doing this kind of stuff. And it looks to me like it's going to both lead to good mechanics, good robotics, interesting problems in control and materials and technology that can be used for a wide variety of purposes. And I'm done on time. Thanks for coming up.