1 00:00:05,260 --> 00:00:09,490 Raul do you teach you're going to do a lot of 2 00:00:09,490 --> 00:00:13,470 Dora work with George Whitesides and I believe happened to be getting off 3 00:00:13,470 --> 00:00:17,710 of moral visionary boots on the terrace which is just a coincidence. 4 00:00:18,890 --> 00:00:25,490 And I get the you are the workshop I think the press all 5 00:00:25,490 --> 00:00:31,970 of this stuff really working on this very thread you got to get another seminar all. 6 00:00:32,980 --> 00:00:37,019 The wealth that Kordell that perseveres 7 00:00:37,019 --> 00:00:41,420 recently in the territory in the US today will work and maybe and so. 8 00:00:42,740 --> 00:00:46,450 Will really go really further what it was on the fourth. 9 00:00:51,850 --> 00:00:52,350 Thanks. 10 00:00:54,450 --> 00:00:59,360 Right and so it's actually Dan's a really inspirational to me because he talks about 11 00:00:59,360 --> 00:01:04,130 robots in the terms of physics and I'm trying to do it in terms of chemistry so 12 00:01:04,130 --> 00:01:06,680 maybe roll chemistry and I kind of steal from that. 13 00:01:08,209 --> 00:01:13,710 But so I want to give it a little historical perspective I entered into 14 00:01:13,710 --> 00:01:19,150 the field of robotics five years ago and then I have to roll chemist and 15 00:01:19,150 --> 00:01:23,070 to really have it was like what's the robot I didn't even know and so 16 00:01:23,070 --> 00:01:27,110 maybe from me telling the story of how I came to understand the fields. 17 00:01:28,600 --> 00:01:32,519 It can and you know maybe I don't know you might be able to learn something from it. 18 00:01:33,559 --> 00:01:35,530 But you don't. 19 00:01:36,980 --> 00:01:38,489 Know it's a better. 20 00:01:41,760 --> 00:01:47,530 OK Right so Matt Cornell and for those of you who are postdocs and 21 00:01:47,530 --> 00:01:51,830 looking for positions I'm spotted leaving a design manufacturing search 22 00:01:51,830 --> 00:01:55,159 there are also two other searches going on under par in our department so 23 00:01:55,159 --> 00:01:58,080 please have a look at it on our website. 24 00:01:59,390 --> 00:02:01,650 A lot of good faculty there they get along with and 25 00:02:01,650 --> 00:02:02,979 we do some I think pretty good work. 26 00:02:04,260 --> 00:02:04,760 So. 27 00:02:05,829 --> 00:02:11,490 Coming from you know just coming new into the field of robotics 28 00:02:11,490 --> 00:02:15,170 I was like OK so what are the challenges and 29 00:02:15,170 --> 00:02:21,010 to me it the field of software box without actually ever explicitly stating it 30 00:02:21,010 --> 00:02:26,379 is trying to help solve the in durance versus agility contradiction 31 00:02:26,379 --> 00:02:30,500 where if you have wheels you know you can travel for a long time. 32 00:02:31,840 --> 00:02:36,819 As a robot if you have legs you're more adaptable but the power 33 00:02:36,819 --> 00:02:40,950 requirements are higher and you're less efficient in using your momentum. 34 00:02:42,069 --> 00:02:46,639 And I thought this this is cool Boston Dynamics handle which is kind of 35 00:02:46,639 --> 00:02:50,450 trying to use wheels and legs but still has obvious issues. 36 00:02:53,711 --> 00:02:59,500 So how biology manages this is through actually using less efficient 37 00:02:59,500 --> 00:03:05,131 actuators but in a distributed way coupled with sensing and local computation. 38 00:03:07,020 --> 00:03:11,391 The reason our robots look like this and not like that. 39 00:03:12,421 --> 00:03:15,271 And conversely why biology doesn't look like this. 40 00:03:16,361 --> 00:03:18,650 One biology can't use metals. 41 00:03:20,190 --> 00:03:26,510 Because they're high temperature processes but biology can use self-assembly to grow. 42 00:03:27,671 --> 00:03:32,541 The distribution of actuators and sensors and neuron neuronal networks for 43 00:03:32,541 --> 00:03:37,251 highly sophisticated adaptation high degrees of freedom and sensory feedback. 44 00:03:39,131 --> 00:03:42,370 The first mention of robot was in Russell of 45 00:03:42,370 --> 00:03:45,771 Universal Robots the really boring play if you've ever read it. 46 00:03:47,891 --> 00:03:53,040 And it just the robot robot in check means drudgery or slave labor and 47 00:03:53,040 --> 00:03:56,220 to me is very interesting that these robots are actually grown they're 48 00:03:56,220 --> 00:04:01,790 synthesized from organic matter is very it's a lot like what I do and 49 00:04:01,790 --> 00:04:04,251 not a lot like what her robots are currently made. 50 00:04:05,891 --> 00:04:10,281 It would be great if we could use self-assembly to totally assemble a robot 51 00:04:10,281 --> 00:04:13,771 and we're not bad at self-assembly synthetically but 52 00:04:13,771 --> 00:04:16,491 certainly not sophisticated enough to make. 53 00:04:17,761 --> 00:04:21,280 A feedback control robot that can do useful work so 54 00:04:21,280 --> 00:04:24,331 I think we need another other approaches. 55 00:04:26,041 --> 00:04:30,310 So I use a true trigger walked in my work milieu were true is. 56 00:04:32,030 --> 00:04:32,791 It is a. 57 00:04:34,920 --> 00:04:41,371 Is a parametric solution space for design contradictions so 58 00:04:41,371 --> 00:04:45,810 if I am connected to the internet I will show you really quickly how to use it so 59 00:04:45,810 --> 00:04:47,101 I didn't see any hands raised. 60 00:04:49,820 --> 00:04:55,079 Well there we go so there's actually a website called Trees forty. 61 00:05:01,460 --> 00:05:01,960 And. 62 00:05:04,019 --> 00:05:07,900 So you can actually enter in your engineering contradictions and 63 00:05:07,900 --> 00:05:14,799 get solutions out this was this is a acronym for a Russian acronym there is a. 64 00:05:16,490 --> 00:05:21,460 Scanning all sure and Soviet Union who went through all the Soviet literature and 65 00:05:21,460 --> 00:05:25,180 found out all the repeating engineering contradictions and 66 00:05:25,180 --> 00:05:31,479 they're repeating solutions the idea being that nothing is new and 67 00:05:31,479 --> 00:05:36,330 which I don't agree with and this process still does require invention but 68 00:05:36,330 --> 00:05:40,740 you can select a feature to improve like the weight of the stationary object but 69 00:05:40,740 --> 00:05:41,720 you want to keep it. 70 00:05:42,999 --> 00:05:45,100 Fast for example and then it'll. 71 00:05:46,790 --> 00:05:48,979 Well this one has no solution proposed. 72 00:05:50,639 --> 00:05:51,940 Anyway you some do. 73 00:05:52,990 --> 00:05:56,150 Well and I'll just go through show you the one that I found. 74 00:06:03,358 --> 00:06:06,919 OK so what I I said is I want to. 75 00:06:08,498 --> 00:06:13,509 Improve the use of energy by a moving object and can serve the adaptability or 76 00:06:13,509 --> 00:06:18,809 versatility of the object and then solutions to that in the dynamic space and 77 00:06:18,809 --> 00:06:23,099 I just took a snapshot there's many more than this if an object or 78 00:06:23,099 --> 00:06:26,388 process is rigid or inflexible make it movable or adaptive. 79 00:06:28,428 --> 00:06:32,468 Allow the characters as an object or process to change to be optimal So 80 00:06:32,468 --> 00:06:37,769 to me this means you know and then you can interpret this solution space to your 81 00:06:37,769 --> 00:06:44,839 particular capabilities what I think it is that we can use shape changing objects and 82 00:06:44,839 --> 00:06:50,239 compliant objects to try and meet that solve this engineering contradiction. 83 00:06:51,979 --> 00:06:54,079 Of those in the first person to think of this of course. 84 00:06:55,189 --> 00:06:59,589 There's examples of extrinsically compliant 85 00:06:59,589 --> 00:07:04,179 robots This is a memory alloy system of course presser we did hears 86 00:07:04,179 --> 00:07:08,179 also works on these systems in the top left with saying big hit MIT 87 00:07:08,179 --> 00:07:12,938 has this thing called a mesh worm which is a heel Ecole shape memory L.-Y. wire and 88 00:07:12,938 --> 00:07:17,089 that moves a robot but it's very compliant so it can sustain damage for example. 89 00:07:18,199 --> 00:07:19,239 That Fukushima. 90 00:07:20,969 --> 00:07:27,018 Has of course an underage under actuated grasper which scared this participant but 91 00:07:27,018 --> 00:07:33,589 presumably she survived and then dollar as these thin polyethylene. 92 00:07:34,609 --> 00:07:35,459 Living hinges. 93 00:07:38,259 --> 00:07:43,839 But what I want to use are polymers you can process is liquids 94 00:07:43,839 --> 00:07:50,389 using chemistry cure them into compliance mechanisms and then ensure 95 00:07:50,389 --> 00:07:54,009 that they're always compliant and the way to ensure that they're always compliant 96 00:07:54,009 --> 00:07:58,859 is by using the last American material very low elastic modulus material. 97 00:08:00,199 --> 00:08:02,909 And this is just a very simple fit. 98 00:08:02,909 --> 00:08:07,859 Non-physical solution gets you understand what they are the top 99 00:08:07,859 --> 00:08:12,969 left is a bunch of noodles that are chemically bonded together and when you 100 00:08:12,969 --> 00:08:16,998 stretch them you're reducing the entropy of the system which is energetically 101 00:08:16,998 --> 00:08:21,959 unfavorable these are high are dominated by entropy the energetically and 102 00:08:21,959 --> 00:08:26,549 so they want to return to their coiled state and so basically by 103 00:08:26,549 --> 00:08:32,399 processing the last immerse and I won't I will go through this is not helpful but 104 00:08:32,399 --> 00:08:36,989 as you stretch them they want to come back so you can pour liquids and 105 00:08:36,989 --> 00:08:44,798 mold springs at the nano scale using rubber This is a video of a. 106 00:08:46,447 --> 00:08:52,168 Leg of the basically a blue that we mold it and so by changing the elastic 107 00:08:52,168 --> 00:08:55,907 modulus of different parts of the balloon we can control it to inflate and bend. 108 00:08:57,378 --> 00:09:01,778 And we're storing energy and that stretched membrane and when we release 109 00:09:01,778 --> 00:09:05,848 the pressure it wants to coil back up the polymers do energetically and 110 00:09:05,848 --> 00:09:09,557 that passively restores the shape of that actuator. 111 00:09:11,148 --> 00:09:13,388 So that's that's an advantage you don't get. 112 00:09:14,508 --> 00:09:20,618 Other mechanisms so it's a can we what can we use this for and how can we use it so 113 00:09:20,618 --> 00:09:25,807 other ways we process materials today like this liquid based processing 114 00:09:25,807 --> 00:09:30,998 there is a rotation casting you can make both holes this way like huge things 115 00:09:30,998 --> 00:09:35,638 injection molding to make things really fast and then three D. 116 00:09:35,638 --> 00:09:39,328 printing and we've pursued all three of these manufacturing techniques. 117 00:09:41,518 --> 00:09:46,417 And I won't go into how each thing is made until later but just assume 118 00:09:46,417 --> 00:09:52,347 normally injection molding so we molded a manipulator 119 00:09:52,347 --> 00:09:58,038 that when it inflates just wraps around an object so there's no feedback 120 00:09:58,038 --> 00:10:01,758 there actually but if you look at it as a material computation there is 121 00:10:01,758 --> 00:10:06,627 because the input to this glove is stress and then the output is strain and 122 00:10:06,627 --> 00:10:10,697 so it's wrapping around this fennel without actually knowing what it is and 123 00:10:10,697 --> 00:10:15,898 we can to know how stiff or how much force these things can apply by changing 124 00:10:15,898 --> 00:10:21,098 the elastic modulus of the material via chemistry if we want it to Delhi can it 125 00:10:21,098 --> 00:10:26,298 delicately manipulate objects will choose a very low lastic modulus material so 126 00:10:26,298 --> 00:10:31,628 the slope of the stress stream curve is low so it's easily inflated and 127 00:10:31,628 --> 00:10:34,937 then less stiff for the same degree of actuation or 128 00:10:34,937 --> 00:10:37,958 we can choose one with a higher stress string curve so 129 00:10:37,958 --> 00:10:41,038 it's harder to inflate higher pressure so stiffer in the end. 130 00:10:46,608 --> 00:10:51,559 So we spent a lot of effort replicating something that shouldn't be replicated 131 00:10:51,559 --> 00:10:52,588 a player piano. 132 00:10:54,619 --> 00:10:58,869 That only plays one tune and that we did that just to show that these things can be 133 00:10:58,869 --> 00:11:03,379 fast a lot of the work in software products before this we're very slow 134 00:11:03,379 --> 00:11:08,149 moving things it's based on the did the pressure differential you're plying it's 135 00:11:08,149 --> 00:11:12,729 not not difficult to understand but all the other prior methods were slow. 136 00:11:13,848 --> 00:11:14,749 OK so. 137 00:11:16,429 --> 00:11:20,878 What I like to do is find a fairly inspirational organism and 138 00:11:20,878 --> 00:11:26,049 then try to mimic some of its features and work with biologists to do it better 139 00:11:26,049 --> 00:11:30,959 than I could on my own and so I work with Roger Hanlon who took this video and 140 00:11:30,959 --> 00:11:35,689 what you can see is a octopus or maybe not see as an octopus and 141 00:11:35,689 --> 00:11:40,849 then what it's doing is changing the posture its color and its texture in order 142 00:11:40,849 --> 00:11:45,878 to effectively camouflage into this algae and the only right way you can actually 143 00:11:45,878 --> 00:11:50,079 see this even after white washing to bunch of times is because it's not moving so 144 00:11:50,079 --> 00:11:54,129 the texture isn't moving but that's the only way you can tell it's an octopus 145 00:11:56,809 --> 00:12:01,868 so what I decided to do is use as an example of what we can do with these 146 00:12:01,868 --> 00:12:06,118 soft materials as mimic these functions the posture color and texture change I've 147 00:12:06,118 --> 00:12:09,679 already shown you the posture change just in the grasp or that's fairly easy. 148 00:12:11,179 --> 00:12:15,239 I chose to mimic bioluminescence 149 00:12:15,239 --> 00:12:19,129 from an octopus which they use for counter illumination. 150 00:12:20,179 --> 00:12:24,258 As a form of camouflaging to do that we embedded light 151 00:12:24,258 --> 00:12:30,109 emitting particles in the dielectric layer of a rubber capacitor and so 152 00:12:30,109 --> 00:12:35,309 this pixel can stretch to over six hundred percent string while 153 00:12:35,309 --> 00:12:39,059 the emitting light and in fact the more you stretch it since this illumination 154 00:12:39,059 --> 00:12:43,469 is feel driven in a sea field it gets brighter as the thickness decreases and 155 00:12:43,469 --> 00:12:47,059 since volume is conserved when you stretch it the thickness is decreasing. 156 00:12:48,719 --> 00:12:52,699 Since this is a capacitor during the stream you can also measure 157 00:12:52,699 --> 00:12:57,668 the strain via the capacitance signal so this is this strain sensor and 158 00:12:57,668 --> 00:12:59,679 light emission capacitor. 159 00:13:01,899 --> 00:13:03,119 You turn it into a. 160 00:13:05,239 --> 00:13:10,709 Into individual addressable pixel array at this time we were using high voltages 161 00:13:10,709 --> 00:13:16,558 like one thousand volts to cause the light emission turns 162 00:13:16,558 --> 00:13:21,469 out we were just really dumb then we could have instead of putting We can actually do 163 00:13:21,469 --> 00:13:25,719 much thinner layers of the dielectric with the light emitting particles so 164 00:13:25,719 --> 00:13:29,429 we can get that voltage down to supper hundred volts but 165 00:13:29,429 --> 00:13:33,388 that's why you're getting these crosstalk because the we make this by hand or 166 00:13:33,388 --> 00:13:38,388 show me this by hand so we're getting at these fields crosstalk between them. 167 00:13:40,138 --> 00:13:44,479 So again so we can use this is a multi touch sensor as well that is an eight by 168 00:13:44,479 --> 00:13:48,369 eight grid of light emitting capacitors that can also sense defamation So 169 00:13:48,369 --> 00:13:51,138 even at two hundred percent aerial strains 170 00:13:51,138 --> 00:13:55,459 we could localize the position of three three touches at least. 171 00:13:56,619 --> 00:13:57,789 And so we're using that. 172 00:13:59,538 --> 00:14:00,038 Sorry. 173 00:14:01,329 --> 00:14:06,419 To control Tetris and so we have so we're feeding that 174 00:14:06,419 --> 00:14:11,529 sixty of those sixty four strain sensors out to convolutional neural network and 175 00:14:11,529 --> 00:14:15,719 then inter-breeding that strain information as gestures and so 176 00:14:15,719 --> 00:14:21,699 we can train this controller to rotate a tetra Mino drop a tetra mino or 177 00:14:21,699 --> 00:14:26,199 you know really what we should have done is worked with amputees and have them 178 00:14:26,199 --> 00:14:32,428 control a computer without a hand but we decided to play Tetris instead. 179 00:14:36,319 --> 00:14:37,428 We get strange we don't. 180 00:14:38,969 --> 00:14:40,448 Get. 181 00:14:40,448 --> 00:14:45,829 What we were getting We don't know which it gives it's the signal changes in either 182 00:14:45,829 --> 00:14:51,518 way so where the the beauty about neural networks is that we don't necessarily have 183 00:14:51,518 --> 00:14:55,739 to know for the application the bad thing is that we also we don't have to know for 184 00:14:55,739 --> 00:14:58,928 the application so we're not learning much in that sense. 185 00:15:00,129 --> 00:15:04,199 So as I said before texture has to change for effective camouflage 186 00:15:04,199 --> 00:15:09,209 this is on a cuttlefish I believe a giant Australian cultish. 187 00:15:10,389 --> 00:15:13,499 And it's pretty remarkable so. 188 00:15:15,299 --> 00:15:19,709 In a very unremarkable way we tried to inflate balloons 189 00:15:19,709 --> 00:15:24,359 into textures that but with apre knowledge so we all three D. 190 00:15:24,359 --> 00:15:29,929 scan an object and then came we came up with an algorithm to take that three D. 191 00:15:29,929 --> 00:15:35,889 shape pattern in extensible rings onto inflatable rubber sheet 192 00:15:35,889 --> 00:15:41,119 if you can constrain the circumstantial inflation of a bloop then 193 00:15:41,119 --> 00:15:45,719 you only get radial or patricians and then it's easier to understand how to pattern 194 00:15:45,719 --> 00:15:51,378 those circles to get protrusions And so using that well known math 195 00:15:51,378 --> 00:15:56,649 James people who's now who was opposed to the time is now fact you Penn came up with 196 00:15:56,649 --> 00:16:00,999 came up with a technique to pattern these rings concentric rings onto 197 00:16:00,999 --> 00:16:05,339 that rubber and inflate it to get either a positive zero or negative Gaussian 198 00:16:05,339 --> 00:16:09,259 curvature isn't in theory you can combine those curvatures to create any shape. 199 00:16:12,069 --> 00:16:12,678 So. 200 00:16:13,849 --> 00:16:18,409 Now what are mature are material can stretch to seven hundred percent strain 201 00:16:18,409 --> 00:16:22,109 but we would need to be able to stretch it to ten thousand percent strain 202 00:16:22,109 --> 00:16:23,869 to replicate most shapes so 203 00:16:23,869 --> 00:16:29,429 what we what we're working on is the subclass of lower aspect ratio shapes that 204 00:16:29,429 --> 00:16:34,608 that we can pattern in that's why we're calling it texture because most textures 205 00:16:34,608 --> 00:16:38,479 at the scale we deal with are are fairly low aspect ratio and 206 00:16:38,479 --> 00:16:44,119 this plant here that we replicated is fairly low aspect ratio if you want to 207 00:16:44,119 --> 00:16:48,388 replicate the plate you all know which is a very tall sharp one with pointy things. 208 00:16:49,638 --> 00:16:53,219 That we would be able to do that with our material but on the other hand you never 209 00:16:53,219 --> 00:16:57,809 actually feel an alibi plant all at once you feel portions of it and so being able 210 00:16:57,809 --> 00:17:02,949 to replicate smaller portions of a shape is what we're going for in this paper. 211 00:17:04,478 --> 00:17:06,829 And this is just showing the inflation of. 212 00:17:08,269 --> 00:17:11,728 Of the bloom into the course approximation of that plant. 213 00:17:18,597 --> 00:17:23,028 OK but soft is not always a good thing I love the octopus but 214 00:17:23,028 --> 00:17:27,248 it doesn't do very well on land the reason of course it doesn't have the point 215 00:17:27,248 --> 00:17:32,458 support of the ocean so when it's on land in order to walk and be stiff it has to 216 00:17:32,458 --> 00:17:36,827 contract musculature all the time whereas we just use our skeleton for 217 00:17:36,827 --> 00:17:39,868 that stiffness so it would be great if we had it so 218 00:17:39,868 --> 00:17:43,998 this is another engineering contradiction that we can use trees to solve. 219 00:17:45,077 --> 00:17:49,687 How do we have adaptability while maintaining strength and so 220 00:17:49,687 --> 00:17:53,687 you can feed that into trees and a solution from trees is actually using. 221 00:17:54,768 --> 00:17:59,918 Face changing materials so how do we apply that to what we do we actually in prior 222 00:17:59,918 --> 00:18:04,848 work had made a horrible lasting foam actuator that you can mold and 223 00:18:04,848 --> 00:18:08,368 then pressurized to Khalid Shaikh changes in three dimensions so 224 00:18:08,368 --> 00:18:12,247 we just filled that Pora lastic material with a little melting temperature metal. 225 00:18:13,437 --> 00:18:17,537 Field metal and melts a sixty three Celsius so it's six yourself see this 226 00:18:17,537 --> 00:18:21,107 is liquid and then you can morph your rubber very well and 227 00:18:21,107 --> 00:18:26,808 then you can lower the temperature below sixty Celsius to walk in a stiff shape so 228 00:18:26,808 --> 00:18:28,528 that's this is it it working. 229 00:18:29,767 --> 00:18:33,678 Silicone fall imbibed with a low melting temperature metal alloy freeze it it's 230 00:18:33,678 --> 00:18:40,168 stiff melted it's adaptable and the capillarity of the interface keeps 231 00:18:40,168 --> 00:18:44,047 that in in that foam structure even though the density difference is quite high. 232 00:18:47,028 --> 00:18:50,788 Change Shape interesting only I will unexpectedly and 233 00:18:50,788 --> 00:18:55,098 I think initially is that the foam pores are randomly arranged so 234 00:18:55,098 --> 00:18:58,517 when you stretch it the pores a line in the direction you're stretching and 235 00:18:58,517 --> 00:19:00,988 then actually it's stronger in the direction you stretched it 236 00:19:00,988 --> 00:19:05,568 in terms of the yield stress now probably gets weaker orthogonally But 237 00:19:05,568 --> 00:19:08,398 you know we didn't we didn't measure that but it think it's a bit similar 238 00:19:08,398 --> 00:19:12,418 to how bone gets stronger on this side you apply more weight to. 239 00:19:15,610 --> 00:19:21,459 And then how this works remember the tropical last to City of the polymer 240 00:19:21,459 --> 00:19:25,990 if you start with a cube and then force it into a cylinder and 241 00:19:25,990 --> 00:19:29,140 then freeze it in that case you're building in stress and 242 00:19:29,140 --> 00:19:32,870 then when you heat it again it will want to react spanned into so 243 00:19:32,870 --> 00:19:38,840 in that case it's a shape memory Al void composite that goes one time 244 00:19:38,840 --> 00:19:44,199 so but if you if you interface that with with an actuator then 245 00:19:44,199 --> 00:19:49,149 it then it could be you could really program reprogram it afterwards. 246 00:19:50,830 --> 00:19:51,840 And that is a hairdryer. 247 00:19:53,039 --> 00:19:54,300 Blowing a hair dryer out it's. 248 00:19:55,550 --> 00:19:57,210 No it is not an autonomy is a robot. 249 00:20:00,360 --> 00:20:05,030 OK So book Why have the last American robots machines not occurred sooner. 250 00:20:06,350 --> 00:20:11,469 That technology is ancient comes from the good a perch a tree 251 00:20:11,469 --> 00:20:15,969 that's natural rubber and the Aztecs used to play a game called Pocket talk 252 00:20:15,969 --> 00:20:17,620 where these tribes would play each other and 253 00:20:17,620 --> 00:20:20,660 then the losing tribe was all slaughtered so you do not want to lose that game. 254 00:20:22,349 --> 00:20:27,599 And but so that rubber ball though was useless as anything else other than maybe 255 00:20:27,599 --> 00:20:33,220 a shoe used to shoes to but definitely as a car tire as a car tire 256 00:20:33,220 --> 00:20:38,410 very low toughness and they just a break and then are ruined after a few miles. 257 00:20:39,459 --> 00:20:43,010 Which became important in World War two when we had all these. 258 00:20:44,060 --> 00:20:49,080 Were machines on wheels and so the allies actually cut off supply of 259 00:20:49,080 --> 00:20:54,349 natural rubber to the axis which really helped us win that war and 260 00:20:54,349 --> 00:20:57,680 then what with that came an effort to create synthetic rubber so 261 00:20:57,680 --> 00:21:01,639 people wouldn't be dependent on this natural rubber supply and 262 00:21:01,639 --> 00:21:06,160 so Charles Goodyear nine hundred forty fell so this relied on natural rubber 263 00:21:06,160 --> 00:21:10,300 Charles Goodyear made the tires the tough tires that everybody used but 264 00:21:10,300 --> 00:21:12,790 there was a huge effort to try to make started be to die in rubber and 265 00:21:12,790 --> 00:21:15,410 other synthetic versions. 266 00:21:15,410 --> 00:21:19,079 Which kind of started working in the late forty's but 267 00:21:19,079 --> 00:21:24,380 then silicone rubber really tough low elastic modulus highly structural material 268 00:21:24,380 --> 00:21:30,390 didn't come around until the mid forty's as well through by work from da Corning 269 00:21:30,390 --> 00:21:33,340 but that made silly putty which is not chemically cross-linked chemically 270 00:21:33,340 --> 00:21:37,920 cross-link version of silly putty which means you can stretch it and come back 271 00:21:37,920 --> 00:21:41,490 didn't come out to the fifty's and then it really didn't start getting used for 272 00:21:41,490 --> 00:21:46,560 anything more than gaskets until the ninety's terms of prosthetic interfaces 273 00:21:46,560 --> 00:21:51,960 and so we didn't really see machines out of this material these types of materials 274 00:21:51,960 --> 00:21:56,510 until the dozen tens when software products incorporated came out and 275 00:21:56,510 --> 00:21:58,430 started selling manipulators. 276 00:21:59,780 --> 00:22:00,420 Fests and 277 00:22:00,420 --> 00:22:05,120 George Whitesides is who is giving a talk here he's a the board of that company. 278 00:22:06,399 --> 00:22:12,250 So and Festo a German company and Empire robotics they all started making these. 279 00:22:13,300 --> 00:22:18,410 Device manipulators soft machines you could buy Empire robotics failed and 280 00:22:18,410 --> 00:22:22,080 there's a nice article from one of the from the C.T.O. 281 00:22:22,080 --> 00:22:26,050 in the journalist software box about this failure the reason and I. 282 00:22:28,080 --> 00:22:31,990 They may all fail because what they're trying to do right now is have 283 00:22:31,990 --> 00:22:37,240 a manipulator for every object rather than picking a target application so 284 00:22:37,240 --> 00:22:40,490 the advantages of robotics is they're highly adaptable but if you do use that as 285 00:22:40,490 --> 00:22:45,000 your value proposition you're unlikely to actually make your company work that way. 286 00:22:46,380 --> 00:22:51,700 So then I've been thinking about what our software bots actually useful for 287 00:22:51,700 --> 00:22:57,029 and try to work in that area and this is work done by Harlow's like college just 288 00:22:57,029 --> 00:23:03,669 showing that monkeys far prefer to be in contact with the soft surrogate 289 00:23:03,669 --> 00:23:08,779 they'll feed from this steel mesh circuit but as soon as you don't 290 00:23:08,779 --> 00:23:12,330 feed in they go back to the software and it's kind of something we all know but. 291 00:23:13,790 --> 00:23:15,640 The this monkey spends. 292 00:23:15,640 --> 00:23:18,549 A thousand times more contact with this software and 293 00:23:18,549 --> 00:23:21,010 the only difference is the mechanical feeling. 294 00:23:23,229 --> 00:23:23,780 So. 295 00:23:24,870 --> 00:23:29,100 It's clear to me that an advantage of software boxes in 296 00:23:29,100 --> 00:23:33,359 Hugh close proximity to humans human robot interactions and so 297 00:23:33,359 --> 00:23:38,769 those are by medical devices have to interfaces perhaps beginning and 298 00:23:38,769 --> 00:23:42,840 end of life care and so we're working on all those spaces 299 00:23:42,840 --> 00:23:46,699 this is an example of a glove that helps assisting grasping for 300 00:23:46,699 --> 00:23:52,140 stroke rebuild patients prosthetics or head replacements. 301 00:23:53,539 --> 00:23:56,960 And then what we did in this case so everything I've shown you before hand is 302 00:23:56,960 --> 00:24:01,960 just a machine it's not a robot you actually need feedback in sensing 303 00:24:01,960 --> 00:24:05,000 the environment and then the response to that and so we've integrated. 304 00:24:07,080 --> 00:24:11,780 A new kind of strain sensor into these into the skins of these robots which 305 00:24:11,780 --> 00:24:16,390 are based on like guides so we just shine a light from one end of like I collect on 306 00:24:16,390 --> 00:24:20,300 the other end and as you stretch it you're increasing the pathlength in three beers 307 00:24:20,300 --> 00:24:22,959 like you get a decrease in power output on the other side. 308 00:24:24,389 --> 00:24:29,170 There's more you can do with that but that's the simplest That's the first 309 00:24:29,170 --> 00:24:34,290 order of what we're using it for and so we mold these light guides and 310 00:24:34,290 --> 00:24:38,940 then we put them in our soft actuators and actually this is 311 00:24:38,940 --> 00:24:43,560 just showing a characterization in other stretchable strain sensors they're usually 312 00:24:43,560 --> 00:24:47,870 conductive So you measure the resistance as they're stretched or capacitance but 313 00:24:47,870 --> 00:24:51,870 those structures will conductors are problematic they're usually filled 314 00:24:51,870 --> 00:24:54,760 with particles that are conductive in a destruction you're rearranging 315 00:24:54,760 --> 00:24:59,630 the configuration of those particles so you get a different resistivity every time 316 00:24:59,630 --> 00:25:04,540 in our system it's the same every time and it's stretches 317 00:25:04,540 --> 00:25:09,130 to just as high a strain or higher than these other systems and is very cheap. 318 00:25:11,150 --> 00:25:13,419 We incorporated that into actually going to go faster. 319 00:25:14,470 --> 00:25:15,170 We incorporated. 320 00:25:15,170 --> 00:25:21,019 All these functions into a single device a hand that can when you pressurized 321 00:25:21,019 --> 00:25:26,410 the hand it acts as a spring as it goes around each tomato so it can talk or 322 00:25:26,410 --> 00:25:32,620 graphically map this table identify different objects and press on them and 323 00:25:32,620 --> 00:25:36,570 measure the force first displacement from each finger so each finger is really. 324 00:25:37,699 --> 00:25:43,210 Mechanical testing machine and then it can select the softest or 325 00:25:43,210 --> 00:25:48,170 lowest stiffness device or fruit which is this tomato or 326 00:25:48,170 --> 00:25:49,879 is a tomato I think tomatoes for. 327 00:25:53,379 --> 00:25:53,920 Me if I'm wrong. 328 00:25:54,950 --> 00:25:58,270 OK And that was that was done by what Chen is now post-doc and 329 00:25:58,270 --> 00:26:03,000 Robin's group at Harvard and we use the same strain optical strain sensors for 330 00:26:03,000 --> 00:26:05,270 the earth thought you can reconstruct the shape of the hand. 331 00:26:06,530 --> 00:26:07,600 For Position control. 332 00:26:09,570 --> 00:26:13,640 So this actually I see a couple people that were at I Corps in San Francisco in 333 00:26:13,640 --> 00:26:18,149 the audience and so this is a presentation my student Brian put together. 334 00:26:19,399 --> 00:26:22,970 Showing some of our work but I'll skip ahead we started working with Nvidia. 335 00:26:24,930 --> 00:26:27,330 To make I have to interface for a virtual reality. 336 00:26:28,420 --> 00:26:33,060 Systems and so we use the most popular controller for these games for 337 00:26:33,060 --> 00:26:34,430 this system for the. 338 00:26:35,819 --> 00:26:40,560 H T C five and then he made a little sleeve that fits over this controller and 339 00:26:40,560 --> 00:26:42,610 then can change the shape of the surface. 340 00:26:44,130 --> 00:26:48,040 And it can do it very fast as well so you can get you can really feel impulses so 341 00:26:48,040 --> 00:26:50,900 you can experience gun recoil for example and stimulation. 342 00:26:52,599 --> 00:26:57,090 And so through this process we talked to a lot of potential what we thought were 343 00:26:57,090 --> 00:27:03,419 customers Sony and Samsung another other people everybody really loves using this. 344 00:27:05,370 --> 00:27:05,989 It does. 345 00:27:07,150 --> 00:27:11,599 Definitely improve the left hand does not have the interface the right one does 346 00:27:11,599 --> 00:27:15,170 on the other hand the right one also has a tether that we need to. 347 00:27:15,170 --> 00:27:17,210 Pneumatically power that that skin. 348 00:27:27,020 --> 00:27:30,100 They they all prefer mostly prefer to use their right hand. 349 00:27:47,899 --> 00:27:52,760 That interface was used ten thousand times maybe three different skins but 350 00:27:52,760 --> 00:27:55,160 it's a very reliable interface coming from a lab. 351 00:27:59,010 --> 00:27:59,510 Where you go. 352 00:28:04,239 --> 00:28:05,809 And you can hear the air compressor here. 353 00:28:07,230 --> 00:28:10,520 So we have a couple of bad things going on we have an air compressor and 354 00:28:10,520 --> 00:28:11,989 we have these these tethers. 355 00:28:13,059 --> 00:28:14,300 I had to wait for the last person. 356 00:28:16,609 --> 00:28:18,629 The last person uses actually the head of the R. 357 00:28:18,629 --> 00:28:23,050 developer for a video for the gamer using which is we are Funhouse That's her and 358 00:28:23,050 --> 00:28:25,569 so she's played this game a million times so and 359 00:28:25,569 --> 00:28:29,130 she really enjoyed that that have to experience so we know people want it. 360 00:28:30,430 --> 00:28:35,139 And we figured there was a market for IT hundred twenty five billion dollar market 361 00:28:35,139 --> 00:28:39,240 according to all the analysis we did and we think that haptics the difference you 362 00:28:39,240 --> 00:28:43,650 know we in a Playstation four is the have to experience even though it's a worse 363 00:28:43,650 --> 00:28:47,889 gaming unit actually and and so you know you if you look at all the controllers 364 00:28:47,889 --> 00:28:52,430 they're basically the same so we figure there's a new there's an opportunity for 365 00:28:52,430 --> 00:28:56,760 controllers so we started expanding on that 366 00:28:56,760 --> 00:29:01,540 experience we had with that sleeve we made and we started three D. 367 00:29:01,540 --> 00:29:06,270 printing these these interfaces this is a projection stereo 368 00:29:06,270 --> 00:29:10,559 printer that uses a continuous liquid interface printing it's very fast 369 00:29:10,559 --> 00:29:14,419 can work with a lot of really good materials like you've injection molded it 370 00:29:14,419 --> 00:29:18,610 this Eiffel Tower which is going to be about eight centimeters tall this 371 00:29:18,610 --> 00:29:19,750 is printed in about six and 372 00:29:19,750 --> 00:29:23,050 a half minutes is one hundred times faster than the other three printers. 373 00:29:24,199 --> 00:29:27,740 And what is even better than that is that material that is printed is very high 374 00:29:27,740 --> 00:29:28,259 quality. 375 00:29:31,130 --> 00:29:32,490 And so we lease a carbon M. 376 00:29:32,490 --> 00:29:37,030 one in my lab which uses that technology and we're using it to print. 377 00:29:38,240 --> 00:29:42,649 Different types of interfaces this is a very simple first passive variable 378 00:29:42,649 --> 00:29:47,410 compliant controller so when it's not pressurized it feels like rubber foam and 379 00:29:47,410 --> 00:29:51,190 when it is pressurized it feels so stiff that you really can't feel the difference 380 00:29:51,190 --> 00:29:56,309 between it would and so you can squeeze on it and 381 00:29:56,309 --> 00:30:00,219 that squeeze will send a pressure signal to control the game and 382 00:30:00,219 --> 00:30:04,010 so this is this is Ben the student using it in a in A V. 383 00:30:04,010 --> 00:30:05,780 our environment that Nvidia made for 384 00:30:05,780 --> 00:30:10,300 us where there's a steel sword in the pressurized stiff case and 385 00:30:10,300 --> 00:30:14,320 then you trigger a depressurization into a very soft phone case and 386 00:30:14,320 --> 00:30:18,240 then it turns into a foam sword so but it's not it's not showing up very well. 387 00:30:19,380 --> 00:30:23,089 There was you know they had so that the takes a little while for 388 00:30:23,089 --> 00:30:23,999 the pressure to leave. 389 00:30:26,999 --> 00:30:31,040 And so what ever and we tried this out at a conference and 390 00:30:31,040 --> 00:30:35,850 everybody really liked it so but that that soft This is still not soft 391 00:30:35,850 --> 00:30:40,060 enough people want it has to be vanishingly soft for 392 00:30:40,060 --> 00:30:45,960 people to actually squeeze on it you can see clearly but people don't want to so 393 00:30:45,960 --> 00:30:49,620 we needed so we needed to actually make our own materials for these systems so 394 00:30:49,620 --> 00:30:52,660 we worked on a very soft silicone that we could three D. 395 00:30:52,660 --> 00:30:55,689 printing using this method which turns out to also be self healing and 396 00:30:55,689 --> 00:31:00,090 we also added electrical functionality to materials as well. 397 00:31:02,060 --> 00:31:05,240 And I'll just play this now so I can go faster later. 398 00:31:09,460 --> 00:31:12,290 So using an AC field you can drive electronics 399 00:31:12,290 --> 00:31:14,500 in these ionic Li conductive hydrogels. 400 00:31:16,029 --> 00:31:18,930 So yeah don't get into we developed. 401 00:31:20,309 --> 00:31:24,289 The filing based chemistry to produce silicones they're highly resilient. 402 00:31:25,330 --> 00:31:29,830 And can stretch a lot and can program the strain in the elastic modulus again very 403 00:31:29,830 --> 00:31:34,880 resilient which is important because we're cycling these high strains systems they'll 404 00:31:34,880 --> 00:31:38,580 break unless you really care about resilience in it and so 405 00:31:38,580 --> 00:31:41,909 this is quite as mascot is touchdown the bear so. 406 00:31:43,639 --> 00:31:47,630 That's it and then we made an actuator out of it and we also made a self healing one. 407 00:31:50,710 --> 00:31:53,110 So we can actually puncture it. 408 00:31:54,139 --> 00:31:57,600 And then sunlight will photo curate and seal it again. 409 00:31:59,639 --> 00:32:03,840 It's all healing so if you use it if you use hydraulics instead of pneumatic that 410 00:32:03,840 --> 00:32:08,930 hydraulic liquid can also participate in the chemistry and healing of the system so 411 00:32:08,930 --> 00:32:14,280 we made an ionic we conducted composite using sulfated silicon nano particles and 412 00:32:14,280 --> 00:32:17,190 try methyl Amien group in the polymer. 413 00:32:18,250 --> 00:32:21,310 Made a very tough hydrogen all your heart your contacts are high bills and 414 00:32:21,310 --> 00:32:25,810 so if you squeeze and pull your hydraulic crap here contact lens it cracks we made 415 00:32:25,810 --> 00:32:29,550 a very tough one the you can stretch a lot and it returns to its original shape no 416 00:32:29,550 --> 00:32:33,480 problem and then also as I showed you before you can drive electronics is 417 00:32:33,480 --> 00:32:38,039 an easy field because you're moving the ions back and forth which is also changing 418 00:32:38,039 --> 00:32:42,710 the electrical contact to the number of electrons on your 419 00:32:42,710 --> 00:32:47,719 electrodes OK so in the future 420 00:32:47,719 --> 00:32:52,769 these are going to actually require multi material integration of these three D. 421 00:32:52,769 --> 00:32:56,489 printed parts to get a lot of the functionality I showed you before 422 00:32:56,489 --> 00:33:00,189 in the laminated cases but in a more sophisticated three dimensional structure 423 00:33:00,189 --> 00:33:03,418 so after doing all this in the high core experience and. 424 00:33:04,629 --> 00:33:07,928 Self Assessment and some customer surveys. 425 00:33:09,159 --> 00:33:14,918 Using conjoint analysis when I determined is that I'll just 426 00:33:14,918 --> 00:33:20,138 just skip this is that we're in conjoint analysis you can have people fill out. 427 00:33:22,228 --> 00:33:26,849 You know example simulated products and then based on what they choose for 428 00:33:26,849 --> 00:33:29,258 the simulated products you can back out utility values for 429 00:33:29,258 --> 00:33:31,728 particular features of those products and 430 00:33:31,728 --> 00:33:35,949 then you can run simulations of different products and so what I did I did that and 431 00:33:35,949 --> 00:33:41,349 with an untethered electric controller that operates four or 432 00:33:41,349 --> 00:33:46,959 five minutes at a time people preferred had a preference over that untethered 433 00:33:46,959 --> 00:33:51,699 pneumatic controller nobody wants to magic but until their pneumatic controller 434 00:33:51,699 --> 00:33:55,799 people still don't want because you need to order supplies for it things like that 435 00:33:55,799 --> 00:34:01,219 tethered Electric which operates per in perpetuity that was people's biggest 436 00:34:01,219 --> 00:34:05,249 preference so what people are telling me and companies are telling me and 437 00:34:05,249 --> 00:34:09,499 basically everybody telling me is that we don't want the magic systems and so what 438 00:34:09,499 --> 00:34:14,538 we've done in the case of pneumatic to be shown what software what's can do now we 439 00:34:14,538 --> 00:34:19,928 need to do is make it work in a more acceptable way using elected electricity. 440 00:34:22,039 --> 00:34:27,580 And this was this is our this is Brian in the blue shirt who is 441 00:34:27,580 --> 00:34:30,940 the leader that I Corps thing in this is that so many where they also said we don't 442 00:34:30,940 --> 00:34:35,989 want pneumatic So now there are some places that would could use a simulated 443 00:34:35,989 --> 00:34:40,900 military simulations automotive design fix places like that but 444 00:34:40,900 --> 00:34:45,009 you know I think that I've kind of done the pneumatic thing and we're trying now 445 00:34:45,009 --> 00:34:49,020 work on other soft systems where that are electrically powered and 446 00:34:49,020 --> 00:34:55,100 a good a good place to start there is intended driven soft actuators 447 00:34:55,100 --> 00:35:00,090 this is a tendon a five sexually six degree of freedom. 448 00:35:01,360 --> 00:35:02,889 And that's that we three D. 449 00:35:02,889 --> 00:35:08,250 printed everything but the but the motors and kevlar water so 450 00:35:08,250 --> 00:35:12,400 it's very fast but it also is very strong so we can. 451 00:35:13,590 --> 00:35:18,060 Crush cans of that for example which you can't can't do in the soft case sometimes 452 00:35:18,060 --> 00:35:21,869 you want to sometimes you don't but another key enabling thing in this printed 453 00:35:21,869 --> 00:35:26,499 hand is that we built in a transmission system. 454 00:35:27,620 --> 00:35:32,600 Out of rubber so we spall that tendon around. 455 00:35:34,010 --> 00:35:38,900 Well around a spool and it starts off with a high radius which means it goes faster 456 00:35:38,900 --> 00:35:42,230 if you close the hand fast but then when a contact something 457 00:35:42,230 --> 00:35:47,109 it pulls that radius inwards decreases you can apply high forces so 458 00:35:47,109 --> 00:35:52,510 this is been an issue in prosthetics you want things to operate quickly but 459 00:35:52,510 --> 00:35:57,210 also apply high forces you don't want to use expensive motors so 460 00:35:57,210 --> 00:35:58,880 all these things you have to get to work together and 461 00:35:58,880 --> 00:36:03,429 be able to print the transmission system in a compact way it really helps with. 462 00:36:04,480 --> 00:36:06,650 That of course you can change the angles of things easily so 463 00:36:06,650 --> 00:36:10,080 that our thumb can actually rotate in words as well. 464 00:36:11,550 --> 00:36:16,080 And this is it these past a transmission systems coming off of the carbon in one 465 00:36:16,080 --> 00:36:20,650 printer and then just you know doing what it does. 466 00:36:23,190 --> 00:36:27,120 OK so the ability of three different tendon routing and combining it with 467 00:36:27,120 --> 00:36:32,239 the prior results I think there's a lot there's a lot of new ways to move forward 468 00:36:32,239 --> 00:36:36,489 particularly in the combination of tendon and fluid systems and 469 00:36:36,489 --> 00:36:41,610 that's that's where we're at now one more thing I'll point out is that any 470 00:36:41,610 --> 00:36:46,370 additional way to solve the problem of fluid systems and 471 00:36:46,370 --> 00:36:52,960 power is to make the fluid the battery so that way the SO in this case there is 472 00:36:52,960 --> 00:36:58,750 an electrolyte in an elite powering a pump that is causing the fluid to pressurize 473 00:36:58,750 --> 00:37:02,380 the actuators which is causing the tail to move back and forth and actually 474 00:37:02,380 --> 00:37:06,070 is more dramatic than it looks if you look from the top it's swimming quite a bit. 475 00:37:07,639 --> 00:37:11,450 So tendons fluids batteries as fluids integrating 476 00:37:11,450 --> 00:37:16,060 all those things together is where we're moving in the future and 477 00:37:16,060 --> 00:37:19,970 so I think it's twelve fifty two so I should probably stop right. 478 00:37:21,560 --> 00:37:23,750 Now again one of the total there. 479 00:37:25,929 --> 00:37:27,280 Was that the only good. 480 00:37:30,440 --> 00:37:32,480 OK look OK. 481 00:37:34,130 --> 00:37:37,210 I'll have to reconstruct a hole then so 482 00:37:37,210 --> 00:37:41,030 I'll just take questions instead yeah I could do it yeah that's all right. 483 00:37:46,960 --> 00:37:49,020 All right well that's a good. 484 00:37:50,490 --> 00:37:50,990 Request. 485 00:37:54,778 --> 00:37:55,278 Are you. 486 00:37:56,359 --> 00:37:56,859 With. 487 00:37:59,319 --> 00:37:59,819 Your. 488 00:38:01,308 --> 00:38:01,808 Bridge. 489 00:38:04,499 --> 00:38:06,988 Robot Yeah I am not actually but 490 00:38:06,988 --> 00:38:09,488 I do know where we're good stuff comes from I'll take it. 491 00:38:11,539 --> 00:38:12,039 In that. 492 00:38:14,028 --> 00:38:14,528 They don't. 493 00:38:15,899 --> 00:38:18,179 Like not like they have never had to. 494 00:38:19,528 --> 00:38:20,028 Like like. 495 00:38:21,128 --> 00:38:22,009 I don't know that probably went. 496 00:38:23,058 --> 00:38:26,229 Through a lot of great mathematicians so yeah. 497 00:38:28,638 --> 00:38:33,569 It's not like this where there's a lot of cost involved in setting up 498 00:38:33,569 --> 00:38:37,298 that infrastructure maybe that's why I don't know what more they get their. 499 00:38:55,779 --> 00:39:00,119 Current through it is actually you know we don't direct current through it. 500 00:39:01,748 --> 00:39:06,438 We oscillate a field around it it's actually very low current So 501 00:39:06,438 --> 00:39:10,289 why don't I don't know we only thing we've done is driven L. 502 00:39:10,289 --> 00:39:12,149 e D. with it but 503 00:39:12,149 --> 00:39:16,789 you could use hot very high voltages if high voltages should work just fine and 504 00:39:16,789 --> 00:39:18,989 low voltages but as far as those so she currently don't. 505 00:39:20,888 --> 00:39:21,388 Yes. 506 00:39:37,409 --> 00:39:41,659 Yeah I would certainly I certainly thought about using this material for 507 00:39:41,659 --> 00:39:43,108 brain computer interfaces. 508 00:39:44,869 --> 00:39:49,159 And there's we've we've tested just three thousand three 509 00:39:49,159 --> 00:39:53,209 fiber blasts which are cancer hybrid almost hard to kill anyway. 510 00:39:53,209 --> 00:39:56,149 But so far those seem to work with them are you talking about for 511 00:39:56,149 --> 00:40:00,799 actually invasive connections or. 512 00:40:04,439 --> 00:40:09,349 So far every the only cell type We've tried is live just fine on it 513 00:40:09,349 --> 00:40:12,589 I don't know if we can work with primary cells that will kill it or 514 00:40:12,589 --> 00:40:16,959 not probably they will have to use some kind of protein with it or 515 00:40:16,959 --> 00:40:20,829 change from a poly chromite base to a poly hollar onic acid base. 516 00:40:32,909 --> 00:40:38,078 OK yeah this is yeah OK I skipped over a lot of these details. 517 00:40:39,828 --> 00:40:43,019 So in my bit from materials we made or 518 00:40:43,019 --> 00:40:46,738 the car the which particular device are you thinking of. 519 00:40:50,438 --> 00:40:55,898 How does a printer work OK sorry did last over that there is a just 520 00:40:55,898 --> 00:41:00,009 like this projector shining an image on the screen you can actually shine an image 521 00:41:00,009 --> 00:41:05,129 through a transparent window and then onto a liquid that turns solid when exposed to 522 00:41:05,129 --> 00:41:10,589 light and would have so this black region would stay until him rise to a liquid and 523 00:41:10,589 --> 00:41:15,689 his white region would turn solid and then agree region would be half way cured and 524 00:41:15,689 --> 00:41:21,189 so you can print one layer at a time you not rostering anything 525 00:41:21,189 --> 00:41:26,439 just boom it's done and then that part sticks to an interface you pull up more 526 00:41:26,439 --> 00:41:30,349 liquid comes underneath and then you shine another image on it in that stick and 527 00:41:30,349 --> 00:41:33,399 so you basically play a movie as it's coming out of that liquid. 528 00:41:36,348 --> 00:41:36,848 As. 529 00:41:40,568 --> 00:41:41,068 Well. 530 00:41:49,169 --> 00:41:49,669 Yes. 531 00:42:14,789 --> 00:42:15,289 I. 532 00:42:20,579 --> 00:42:22,359 I see more thought X. 533 00:42:22,359 --> 00:42:24,638 and not just prosthetics or eating but taking. 534 00:42:25,868 --> 00:42:28,469 A semi functioning hand in improving its. 535 00:42:41,289 --> 00:42:45,828 Style that ensued existing prosthetics and overlay a sauce on top of it. 536 00:42:46,839 --> 00:42:47,619 We have not thought of that. 537 00:42:48,749 --> 00:42:50,098 There might be advantage that. 538 00:42:53,089 --> 00:42:53,589 One. 539 00:43:10,219 --> 00:43:14,178 It's a soon how the conjoint survey worked is we created 540 00:43:14,178 --> 00:43:19,119 a simulated product and then changed different. 541 00:43:20,358 --> 00:43:23,409 Tried to gain understand consumer preferences 542 00:43:23,409 --> 00:43:25,928 over different things we could change about their product. 543 00:43:27,219 --> 00:43:32,369 So we gave an example of a controller that looks like a cell phone but 544 00:43:32,369 --> 00:43:34,239 shaped like a cell phone but 545 00:43:34,239 --> 00:43:38,559 you can change the color the feel whether it can cause a statically changed shape or 546 00:43:38,559 --> 00:43:42,119 kinesthetically change it whether use electrical power or 547 00:43:42,119 --> 00:43:46,689 pneumatic power if it is pneumatic power you have to replace you have to replace 548 00:43:46,689 --> 00:43:52,059 the power source every day and so these different. 549 00:43:53,279 --> 00:43:55,919 Options I'm not sure I'm not answering question I don't think. 550 00:43:59,929 --> 00:44:01,508 You know you know. 551 00:44:03,508 --> 00:44:06,338 If for these surveys is very critical you try and 552 00:44:06,338 --> 00:44:11,659 convey the what the reality of the experience but through pictures and 553 00:44:11,659 --> 00:44:16,269 description I sent I gave a link for the survey but to that that maybe so 554 00:44:16,269 --> 00:44:19,529 maybe you could all fill it out but I didn't send it out in time for 555 00:44:19,529 --> 00:44:24,189 y'all to get it to them but if you want to see it maybe they can send it out later. 556 00:44:32,249 --> 00:44:32,749 If you.