About : standard furniture panel mirror setup
Title : standard furniture panel mirror setup
standard furniture panel mirror setup
we called it 4d printingbecause we wanted to add time. and with stratasysand autodesk, we started to develop a workflow. so the first is theprinting capability. we use multi-material printers. in this case, it'stwo materials. one is a rigid polymer. and like i said,that's the structure. it's the angles.
it's the joints. it's the backbones. the other is an expandingpolymer that expands then, allowing it to transform. and depending on how muchyou put and where you put it, you can control the mechanisms. i'll show two that ithink are promising and more recent examples. so this one is a flat sheet.
uses differential expansion andcontraction across a surface to get positive andnegative curvature. so we were able to makethis mechanism that allows us to controlexactly how much it's going to expand in every location. and therefore you cancontrol how the flat sheet turns into curvature. and obviously, there's someinterest in terms of apparel, and curved panels forautomotive and aviation,
et cetera, instead offorming techniques. and the other one that ithink is quite interesting is how we can think about--this relates to yesterday's discussion aboutdesign, the solution emerging-- how can we thinkabout producing materials that the end resultis not predetermined and we're just trying to get itto transform to the end result. but rather, it dependson its environment. it depends on how longit's in that environment,
it depends on otherfactors, how we activate it, et cetera that differentsolutions can emerge. this is the samepiece at the bottom, drastically different solutionsgiven the amount of time it's in the water. so you have agradient of material. more expansion materialtowards the outside, less towards the center. if it's in thewater for an hour,
you get the middleone-- or no, 15 minutes, i think, to the middleone, an hour on the right. so you get differenttypes of structures. we're interested inpushing that further. but the fundamental questionthat many people had was, how do we program othermaterials than just printed plastic parts. can we do this with carbonfiber, and wood, and textiles, and rubbers, and metals?
and we think the answeris definitely yes. and we're tryingto broaden this. so it's not just aboutprinting anymore. there's many, many capabilities,many material processes. it's really about how toprogram them to transform in some fundamental way. and this grew intoan amazing team. and we're still working on this. and you saw some of ityesterday from christophe.
christophe's a productdesigner and came here, as you heard, with erik demaine. erik obviously will talk later. carbitex, autodesk-- it's agrowing team of amazing people. we've released threematerials so far, or three techniques, carbonfiber, textiles, and wood. i'll show the wood workquickly, because it wasn't touched on yesterday. so there's a longhistory of using wood
as an active material,from japanese joinery, to eames furniture,to achim menges at the university of stuttgart. i think achim's in theaudience somewhere. and there's a lot of workthat has gone into that. i see two main challenges. one is the amount of energyput into forming these chairs, let's say. so you have molds, youhave steam chambers,
and you're forcingit into place. but on the right-handside, we're constrained by existinggrain direction. so the example ithink of is if we want to make mitfold, we don't want to have to go tothe forest and find a tree that says mit in it. so what we starteddoing was printing wood in collaboration with achimand university of stuttgart.
basically, you takewood particles, you can create a fiber from it. there's also off the shelfwood that you can print with. and then you print the grain. so you can create curved grains. you can create convergingand diverging grains. you can create anygrain you want. you submerge it in water. and then it transforms.
you can also use amoisture rich environment. you're basically using themoisture to expand the fibers. you've printed the structure. there's two differentgrain directions. you can do curling. you can do folding and twisting. you can do positive and negativecurvature and eames elephants, anything you can think of. one of the challenges thatwe're interested in exploring
is scalability interms of printed wood. but i think it opens up anew possibility for consumer products and, obviously,wood-based products. this, i wanted to show,is about the textile work that you saw christophetalk about yesterday. and we just showed this atthe milan furniture fair. so this is really scaling up. and you can just bump it. you can really just pull on itor trigger it to have the face
transition from flat. and it jumps into 3d space. and a grown humanbeing can stand on it. it's structural inits second state. so that, i think,shows where we're going, that it's notjust about printing. printing is awesome. printing allows multi materials. there are a lot of capabilities.
we've heard a lot about it. but there are other ways interms of lamination, weaving, knitting, injecting, et cetera. so this is our active textile. that one we callpassive-active, meaning it jumps, the furniture piece. this is heat active. so it transformsbased on temperature. you saw some of thecarbon fiber work
yesterday where we print activematerials on top of the carbon fiber to allow it to transform. we can get folding,curling, twisting. depending on theamount of material we put, we get different forces. depending on thegrain direction, you get differentmechanical behaviors. depending on thematerial property, you can get heat activation,water activation,
light activation, et cetera. and so this has become afull suite of materials with different mechanisms toget precise transformation, control over angles, controlover bistable or single stable. so you saw christophefrom airbus, yesterday, talking about thebistable mechanism. so we can control that. but more related to thetheme of this session, the other category ofwork that we do at the lab
is focused on self-assembly. and one of the firstcollaborations we had was with arthur olson. and you'll hear him speak next. we did this at ted. this was 500 ofthese glass flasks. and in each flask wasa different structure, had different numbersof components. this one is basedon the polio virus.
and basically, you shakeit hard, it breaks apart. you shake it a bit softerbut still randomly, and it comes back together. in art's lab-- ithink he'll show more of this kind ofwork-- has been working on this for a long time. and it really shows, to me,the fundamental principles of self-assembly, thatrandom human interaction or random energy input canlead to non-random results.
you don't need to know aboutthe construction process necessarily. but rather, you need tocontrol the amount of energy, control the ingredients,their interactions, et cetera. we worked on amuch larger version of that where we weretrying to, again, make these furniture pieces. and the shift herewas that we didn't want to just buildan installation,
but rather, an insulationthat produces installations. or we wanted to buildthis machine that would self-assemble the furniture. so we called it theself-assembly line. and people comeup and add energy. so you stand there, andyou rotate this thing, and you get tired. and it's not aseasy as it sounds. and probably, humanismwas better in this case.
but it was reallya demonstration that it is potentially scalable,that it can be much larger. and maybe more than that, itbecame a discussion piece. because people fromdifferent domains came together to talk aboutthis very non-intuitive process. and that's maybe onethread of the work that i'm interested inis changing the intuition for non-intuitive things. like, can we utilizephenomenon that
may happen at another scale? can we utilize that for designand products, architecture, construction, manufacturing? this is a process that wedon't use fundamentally at the human scale. and so it's very interesting tothink about other possibilities for materials comingtogether et cetera. but then we startedto look at how we can go intoself-organization--
so, the same fundamentalcomponent that can come together many different ways. instead of always comingtogether to be the same thing, what if we used the constructionprocess as a design process? and you heard aboutthe silk polymorphism. and that's really oneof the fundamental goals here, that it has thispolymorphic capability to go to differentstates like sheets and lattices, one-dimensionalchains, polygons, et cetera.
so this is thefirst project where we did a large tank of water. we have hundredsof these objects. they're neutrally buoyant,so they move freely in all three dimensions. we can control the turbulence. you see these blackpumps in the back. and we can increase it to blastthem and break them all apart. we can decrease it.
we can have differentpatterns and oscillations. and this allows us to controlboth the location of that-- we can move them--and the intensity and look at howdifferent patterns form, both local patternsand global patterns. so locally, you'll seepentagons, hexagons. you'll see linear strands,sometimes more polyhedra type shapes. other times you can look at itglobally and just see density.
you can see clumping. if you bring the pumpsall the way down, they'll clump into a largemass, much like a solid. and then if you blastthem all, they'll be much more like a gas. they'll break all apart. so this was the first projectentering into that space of fluidic self-organization. but one of theconstraints that we had
was that there was the monkeyand the typewriter challenge. like, if something showedup, would we even know it? how long would it take? how do you recognizesomething shows up? and what happens if it does? can even use it? because every time we would takesomething out, it falls apart. so we then startedto look at lattices. lattices could both bevisually recognizable--
so we could track it. we could visually see it-- andthey would also be more stable given the geometry. so we could pullit out, potentially use is for something, breakit apart, and put it back in. this was also incollaboration with art and with neil gershenfeld's lab. you saw yesterday, neil wasdoing a lot of lattice work with the digital materials.
and so we were subtly tryingto hint that maybe there's another way to assemble theselattices rather than fully robotic assembly. so we went the other end ofthe spectrum where there's no robotic assembly,no human assembly, but rather, depositthese materials-- again, neutrally buoyant. we got a bigger tank. it's 500 gallons.
and they assemble intothis single crystal, a cubic lattice. this video took seven hours. we shot it overnight. and now we can get itroughly in an hour. so what we do now iswe blast the energy. so we would put them all in. we just mix them up as muchas possible, blast the energy, and then cool it off--so, dial it way down.
and very quickly, theyaccumulate into the middle, into a single crystal. well, quickly-- an hour. but we're interestedin scalability. how big can we build it? how many can we build? how fast? what are the patternsof energy-- so, can we create a vortex tocreate towers or linear lift
to create sheets, et cetera? this project looked notat the aggregation and not every component the same. here every componentis different, which is the firsttime that we did that. and every componenthas only one spot that it can go intothe final structure. so it has to findexactly the right place to come together to makethis really ugly chair.
christophe was at thelab just hating us that we make thishideous, hideous chair. but it's quite aninteresting one for us, because the differentiatedcomponents can make this final precise structure. then we went larger. so this project was lookingat 36-inch diameter weather balloons with thesefiberglass frames that float around in thecourtyard outside our lab.
so our lab, thewindow's on the left. and they float around. we would throw a party. and people would come--we did this twice-- and push all the balloons up. we had big fans as well, tryingto create the turbulence. and you'll see that theyslowly start to come together. there's differentlocal connections here. one big breakthroughthat we had here
was that it'svelcro, not magnets. and that was aninteresting step forward. because we then are lookingat other modes of connection and really now startingto think about it as, it's any type of bonding. but you want weaklocal bonds that can error correct, breakoff, that the structure can fix itself. so we're now looking othertypes of mechanical bonding,
and velcro, and magnets,and all different things so that maybe it's more scalablein different environments. one of the challengeshere-- and we're still trying to continue thiswork with marcello-- actually, connectionto the badge. we're interested in makingthese things communicate and having differentstructures light up, different colors, et cetera. and one of thechallenges of that is you
can't leave it for seven hours. you can't leave itfor weeks on end. so you're trying to findways that you can quickly learn and get feedback. and with self-assemblythat takes a while to build this intuition orbuild the right parameters, the right environment. but we're proposing that youcan make large space frame structures that you canhold in your fingertips.
and they come down and you'releft with this large space frame after the balloons die. so you've seen a few imageslike this over the conference where they have this car. and you have lots ofparts disassembled. and to me, it's reallyindicative of where we are right now in termsof human-scale construction. we have lots andlots of components. they all need to cometogether in very precise ways.
so the holy grail solution isthrow a lot of robots at it or a lot of skilled labor. and we're trying to just changethat mentality that there might be other ways, that wecan integrate humans, robotic assembly,and material assembly to rethink how materials cometogether to potentially build more precise structures, tobuild smarter structures, more transformative structures. so it's not just aboutrobots can do everything--
they can assembleeverything, put a lot of motors andsensors and everything-- but rather a new typeof robot, a softer, a more agile, a materialtransformative robot that can both potentially assembleitself as well as transform after it's built. andyesterday i hinted at this idea that we've been toying withwith evolutionary fabrication, this idea that we can fabricatethings using self-assembly or directed self-assembly,fundamental parts that come
together to build structures. but they come together based onthe factors of the environment. so if you're trackingthat and you're computing possible solutions,you're weeding off branches, you can then updatethe pattern of energy. you can update certainfactors, potentially update the shape itselfor the interaction, and potentially buildthese optimized structures. it might be-- thesolution can emerge.
the final solution is muchlike topology optimization in software, but aphysical version, that it can reconfigure tofind a better solution given the scenario. so i'll leave you with this. i think this quotesums up our vision and fits well with theagenda yesterday and today. we believe it's "now possibleto program nearly every material to assemble itself andtransform in useful ways."
thanks so much. [applause] thank you, skylar. let's welcome our nextspeaker, arthur olson. he's a professor ofmolecular biology and director of the moleculargraphics lab at the scripps research institute. ok. well, it's a realpleasure to be here.
this is an unusualmeeting for me, because usually igo to meetings that have mostly structuralmolecular biologists. and here i'm learningso much about how people are tryingto take new concepts and make new thingsthat actually are active in one way or another. oh. oh, sorry.
and so i come from the other wayaround, which is that, in fact, the best example of activematter are you guys. biology has made each one of usable to put ourselves together in some amazing way. and there are a lot a lessonsand a lot of concepts, i think, that are important in tryingto either use that technology to engineer it in some wayor to mimic it in some way. so i like to think of theage that we currently live in as the bio-atomic age,which means that we now
have a conceptionof living things as a physical continuity fromthe atom to the organism. it doesn't mean that weunderstand all of it. but we don't think thatthere's any missing magical physical forces orchemical reactions that need to be discovered,in some sense, in order to explain thebiological phenomenon. but the complexityof the study is such that physics, on one end,is typically formulaic.
you can write it out asa mathematical equation. the complexity that chemistryadds is combinatorics. you just take, let's say,the six most common elements that we're made out of and makesmall molecules synthetically. and that universe of compoundsis 10 to the 60th power. there are 10 to the60th different molecules that you can make, potentially,that contain the elements that we're made out of. and that's more than, ithink, the number of stars
in the universe, at least. so the complexity is very highin terms of the combinatorics. but in fact, evolutionchanges these molecules in dramatic ways. because the molecules thatexist in biology have evolved. and so they're not justgo from the bottom up, they're built from the top down. in other words,these molecules have responded to theenvironment in a certain way
such that they got selectedas part of this living thing. and we have different--so essentially, my lab is computational. i have to make an apology. i don't make anything. the only thing i makeare visualizations. some of them happento have physical form. but i'm in awe ofpeople that make things. so i appreciate every oneof you that has spoken.
but we have different kinds ofmodels, computational models and physical models, at eachof these levels of complexity. physics, as i said,uses analytical models. chemistry adds thestatistics of mass action to those kinds of modelsbecause of the combinatorics. and biology is--well, evolution i like to think of asbeing scale-free. evolution happens at all scales. and every scale affectsthe other scales.
so you need different kindsof modeling techniques to model at thesedifferent scales. so biology requiresmulti-scale, multi-mode models. and that's some ofthe things that i think are the importantareas of research that are going on today. just to give youa sense of size, this is a zoom, with apologiesto charles and ray eames, who did the originalpowers of ten.
and this is not mine. this is from theuniversity of utah. but basically, you startwith a coffee bean, which is kind of the flyingthings that we saw yesterday. and you go down 12orders of magnitude. and now you get tobiological molecules, individual amino acids,and the carbon atom. so as feynman said, there'sa lot of space down there. and the scale is important.
and i'll get tothat in a moment. from a chemicalpoint of view, you can think of life as beinga synthetic technology in some sense. so what are the tools that wehave for synthesis or building large scale molecularstructures from the bottom up? well, first there'schemical synthesis, which people are great at. but the larger the molecule,the more complex the synthesis.
the other thing is thatchemical synthesis involves covalent bond formation. and that's a highenergy process. it's 60 or morekilocalories per mole. and so that haslimiting effects. to make larger moleculesnot through just a complex synthesis but througha polymerization reaction where all of the elements areidentical in one sense, you can make verylarge molecules.
but they tend to besimple chemical subunits. but they're quite valuableas far as our clothing and everything else. i mean, we all basically livein a sea of polymers these days. but there are otherkinds of organization that are higher order. so for instance, these longchains or some other kinds of things can assemble togetherin an organizing association based on some kind of looseaffinity like hydrophobicity
and hydrophilicity-- sothings like oil and water, so you get phase separation. but you get weak bondsbetween the members of the organization. and typically you'll get what'scalled a phase separation. two or more things separate outinto identical organized areas. so the thing about thiskind of association is that it can be very useful,but exact molecular positioning is not known.
it's a statisticalmix of interactions. and finally, you can havenon-covalent molecular assembly, self-assembly. and in that case the idea, asyou saw from skylar's talk, is that you get specificinteractions and specific kinds of structures. there can bevariation, of course. in the biological world thebuilding material is proteins. and you can think of proteinsas self-describing machines.
in other words,they've been coded to fold up into somethingthat has a function. so their linearsequence basically dictates their shape, whichdictates their function. and they can be a varietyof different sizes. and they can assemble. this is a poster that davidgoodsell, my colleague, did. and you can go from verysmall proteins, which tend to diffuse rather rapidlyin the bloodstream-- so things
like insulin andso forth-- to very large assemblies ofmany, many proteins like this microtubulin. this titan is a single molecule,but a very, very long one. and it's kind of likethe tensile member in muscle in some sense. and so you have all kindsof shapes and sizes. and each one ofthese things does one or more specific functions.
so size and time scale--very interesting. you know, some things scale withlinear dimension or geometry. other things do notscale with geometry. the forces between things,the different kinds of forces, the distancesbetween two things, and the falloff of thoseforces does not scale the same as geometry, necessarily. in fact, skylar andi found that out when this small self-assemblingsystem blew up
to the largeself-assembling system. the strength of themagnets had to go up much more than linearly. the magnetic falloffis much shorter than the geometric scaling. so in terms of the sizes,we're talking about very, very small things, as you saw. you're in the sub-nanometerto nanometer range. and so you don't have justone particle coming together
with another particle. you have concentrationsof particles interacting with concentrations of otherparticles in a random fashion. but it's because youhave this mass action. you've got 10 tothe 23rd molecules per mole of a compound. so even if you'remicromolar or nanomolar, you still have a large numberof molecules interacting. and the time scalesare very, very fast.
you're talking about femtosecondto picosecond and nanosecond timescales. so a lot of things can happenin human time perception. ok? protein folds in under a second. and that's a verycomplicated thing. the other thing is thatmolecules exist-- first of all, they exist in water. water is the most importantbiological molecule
that there is, becauseeverything else is a response to water. and this water is thatambient temperature, which means the molecules aremoving around in a random way called brownian motion. so things movearound by diffusion. and they can be impeded, or not,by the hydrodynamics of motion due to the shapes, for instance,of the molecules and so forth. so these are the kindof physical things
that happen at this scale. and as i said before, itis a stochastic system. let's see. i better-- so i won'tdwell on these things. but the point is that asingle protein folds up based on the sequenceof amino acids and the very weak interactionsthat go on between them. these are hydrogenbond interactions that are on the order ofthree kilocalories per mole,
not the covalent interactions,which are 60 or so. so this thing folds up, butit's a knife edge kind of thing. lots of small interactionscause folding. and then the shapesallow them to interact with other proteins. so these are representationsof proteins in the polio virus. and so these shapescan, in fact, be like bricks that canform larger assemblies. so this forms a perfectpentagonal assembly.
and 12 of those cometogether to form the virus. so a lot of thingsare important here. one of them is,in fact, symmetry. this is a symmetric object. and so symmetry doesa number of things. first of all, it allows youto use the same chemical, biochemical, overand over again. and so that savesgenetic information. you just make more copies,higher copy numbers
of structural proteins. they can be long tubes,or they can be in planes, or they can be inthree dimensions. so the symmetrythat we see, these are the point symmetry groups,which are closed groups. all the other ones need somemechanism to stop growth. like crystal growthor fiber growth, there needs to be a mechanismthat stops that growth. with point groupsyou don't have that,
because it's a closed system. and the largestsymmetry that you can have for identicalsubunits is the icosahedron, where 60 identical proteinscan come together and form the viral capsid. this this is a moviethat i did back in 1981. i thought there shouldbe a little history here about graphics. and this is when we solvedthe first virus structure,
just down the road at harvard. and the idea herewas there there were 180 proteins in this capsid. so how did you get 180with icosahedral symmetry? and hearkening backto buckminster fuller and frequencies ofdomes and so forth, it depends on how many threefoldsor sixfolds and fivefolds that you have. but each of these proteins,even though they're chemically
identical, are indifferent environments. they're in differentsymmetric environments. some pack aroundthe fivefold axis. some pack around thethreefold axis here. and some pack aroundthe twofold axis. so those are all the symmetryaxes of the icosahedron. and each one of theseis, in fact, identical chemically, but has shiftedor changed in a certain way so that it couldaccommodate and assemble.
so there's a mechanismfor symmetry. and there's a mechanismto break symmetry but keep some symmetryas part of a component. and so i won't go into theenergetics in any detail. but in fact, thereare two components. one of them is howstrongly two things are attracted to each other. the other one is how much orderor disorder is created when these things come together.
in other words, thisis the entropic term. and we all know that thingstend to get more disordered. and entropy is apenalty if you're freezing out any motions. so anyway, this is exactly the-- [video playback] -utilizing solidprinted models-- ---same thing that yousaw, in fact, skylar. ---of virus components--
-this is on youtube. i put it on, i think, 2007. ---and magnets representing theattractive forces between them, we can produce aphysical analog-- -so this is the self-assembly. ---capable of self-assemblyby random shaking. the model demonstrates notonly how nature accomplishes this feat, but can show someof the subtle characteristics of the stochastic assembly.
notice that thesubassemblies form and break apart en route tothe most stable structure. -the strength of the shakingis analogous to the temperature of the molecular environment. if it's too hot, thevirus can break apart. if it is in theproper range, assembly will take place atan optimal pace. [end playback] so as you saw,this thing actually
assembles pretty quickly afteri said that things do not scale. so the question is, whydoes it assemble so quickly. and the answeris, in some sense, it's the virtualconcentration in that bottle. there are-- from thecombinatorial point of view, there are 11 factorial times5 to the 11th different ways that those piecescan come together. so you're not making a uniqueobject, you're making one of 10 to the 15th objects.
so since i'm runningout of time-- [motor running] ---this is a project thatskylar and i did together. [pieces shaking] so there's two things. one of them is that there's afunction of concentration here that exists. in other words, themore pieces you have, the more reactions you have.
once things startto form, you're getting reactionproduct limited. and at the end, you see thatnot all of them come together. and the same is true, forinstance, for those dna origami experiments. it's a mass action thing. and you don't get 100% assembly. that's not the waybiological assembly works. i was going to talk about motionand maturation of viruses,
but i don't want to gointo too much detail. the fact is that this virushas a lot of subunits. and in fact, it maturesto accommodate the dna as it comes in. this is a virus of a bacteria. it's called a phage. and the interestingthing about that is not only the motion ofthe individual subunits, but in fact, that there'sa topological change that
takes place. so these are actuallyparts of the protein chain. and molecules arebrought together that want to react ifthey're close enough. and that's what happens. and the interestingthing about it is that thiscreates a chainmail. so this virus can actuallyhold 60 psi internal pressure without any problem.
so i will just skipto the end, because i have to show this one thing. so we're talkingabout active matter and how complexliving things are. and so this is thismolecular motor that i was talkingabout called kinesin. and it walks alongthe microtubule. and it doesn't have a powerstroke like people talk. it's a ratcheting device.
it moves around basedon brownian motion. and then when atp comes in,it changes the confirmation. and it locks it in a certain wayto accommodate an interaction when that happens. and then there's a reaction. atp, as it'shydrolyzed, goes to adp. and then it releases again. so it's a simplemolecular system. there are two systems there.
there's the tubulin thatassembles into these railroad tracks. and then there'sthe kinesin, which walks along the tubulesin a particular direction. now all you have to do is toglue a bunch of these tubulin molecules together likea star, where you have a bunch of motors sticking out. so that's another proteincalled streptavidin. and you throw thisinto a container.
you also throw ina crowding agent, something that kindof mimics the density of a cellular environment. and you start toget a phenomenon where the kinesinsthat are walking in different directions--or the tubulins which are oriented in differentdirections-- extend. and then they reach acertain buckling length. and then they collapse.
and then they start again. so what happens is youhave to add atp to this. otherwise it's just kindof a pneumatic mixture. but then you getthis kind of motion. so now you can throwthat into a phase system where you have water inoil or oil and water. and you start to see this kindof dynamic pneumatic phase working around on thesurface of the droplet. and it works for large droplets.
but if the droplets are smallerthan the buckling length it doesn't work. so you get thisidea that you can have-- these are cellularscale objects that are moving around freelywhen you feed it with atp. and this is the lastthing i want to show, which is if you put thisinto a lipid membrane and basically reduce thesurface to volume ratio, you get these tubes stickingout and moving, or swimming,
in certain ways. so the point of it is that withfour simple evolved molecules, you can make dynamic systems. this is a totallyartificial system. and each one of thoseproteins, for instance, can be engineeredin different ways. so you can start to buildfrom the ground up using the fundamentalprinciples of biology. and i'll stop there.
sorry for going over. thank you. i would like tothink arthur olson. we would like to hear more. but unfortunately, yeah,we need to run along. so let's welcome ournext speaker, john main. he's a program managerin the defenses science office at darpa. i have two observationsfrom yesterday,
and i'm here to givea couple of both. one is i think--i've been following materials science for,probably, about 25 years now. i've been following materialsscience for about 25 years now. and i can't think of a timewhen it looked more like chaos than it does right now. and i think that'sa really good thing. i think these arethe kinds of times when disciplines get recast.
and lots of gooddiscoveries get made. and perhaps, new products andnew processes get implemented. the second observationis, this event sort of feels like one of thoseseminal events that's going to help recastmaterials science. so i hope that comes to pass. i actually went backto my room last night and threw away about2/3 of my charts. because after listening to allthe presentations yesterday--
and darpa wasmentioned so many times that i felt like it wasimportant for me to come back and just kind of describethe way darpa thinks, right, or at least my way onthe way darpa thinks. because i thinkit's very possible that a lot of folksthat are in this room might want to turn around andhave a conversation with darpa at some point. and so i'm going towork through that
and hopefully give you abetter idea of how darpa thinks, how to approach darpa. and maybe we can do somegreat things together. it's impossible to have aconversation about darpa without talking aboutwhy darpa exists. and this is exactlywhy darpa exists. darpa exists because,in the late '50s, the russians launched sputnik. it was a huge surprise.
we had no earthly idea that theyhad that kind of capability. and lots of thingschanged after this event. darpa was createdbecause of this event. nasa was created prettymuch because of this event. in fact, the firstdarpa program-- one of the first bigdarpa programs-- was the saturn v launchvehicle that was subsequently transitioned to nasa. so darpa's role, as i tryto tell people all the time,
is we're not a funding agency,although we fund things. we're not an s&tagency, although we work a lot in scienceand technology. fundamentally, whatwe're trying to do is prevent and createstrategic surprise. that's what we do. if we could do it other waysthan fund s&t, we would do it. and we do occasionally dothings other than fund s&t. but the calculation that we're--people bring us great ideas all
the time. and they are invariablydisappointed if we don't throw lots of money at them, right? we all hate getting-- i'ma recovering academic. and so we are allincredibly disappointed when our brilliantproposal didn't get funded. but darpa's calculation'sa little bit different than the nsfcalculation of the nih calculation.
and just to nerd outfor just a second, this is kind of whatdarpa is trying to do. we're trying to figureout if the probability that your greattechnical idea is going to turn into astrategic surprise is much, muchgreater than chance. that's really whatwe're trying to do. so you have the besttechnical idea in the world. but if on examination,we don't think
it's going to turninto a revolution or a strategic surprise,you're probably not going to get funded by darpa. so this is reallywhat we're trying to look for when we evaluateyour proposal, is, is there a strategic surprise. is there a revolution in there? and it's not bad ifthe answer is no. but it's not a darpaproblem if the answer is no.
so darpa has a long, longhistory with materials science. and this is one ofour internal charts that we like to throwaround and explain all the wonderfulthings we've done. and darpa's funded alot of important things over the years. and i'm not goingto go through these. but what i'd like to seeon the right-hand side of this chart in about fiveyears, or maybe 10 years,
is about five pictures thatgrow out of this meeting, of active matterprojects that turn into the seminalsorts of advancements that darpa likes to claim. you'll also notice thatmost of these things on here are very appliedkinds of technologies. you're talking about telescopemirrors, and aircraft parts, and things like that. so it's important to understandwhat space darpa lives in.
there's a good book thatcame out about 15 years ago. and i like to reference this,because i think it clearly explains what the defensesciences office does, which is where i live, iswhere we live in the world, in the s&t world. we don't live in the purebasic quadrant called the bohr quadrant. because a good idea forthe sake of the good idea is really not what we'retrying to accomplish.
there are agencies thatfund those sorts of things. and universities actuallyinternally encourage that kind of thing. and we think it'sterrific, because that's one of the quadrantswe like to mine. we don't live in the pureapplied research quadrant either, just building-- we'renot a product development house. where we live is in thatupper right-hand quadrant,
the use-inspired basic research. we're typically tryingto really mine, right? we're trying to sift through thebasic research that's going on in the world and look forapplications that might turn into strategic surprise. so this is really important. it's important to be able totie the work that you're doing-- if you're seekingdarpa supporter or seeking darpa funding--important to be able to tie it
to what uses are going toemerge from that technology. and that's just anabsolutely irreplaceable item in a darpa proposal. so as far as activematter goes in the talks that we've been listeningto today and yesterday, we have one programright now that is pretty much directlyin the realm of, i think, what active matter is going for. and that's the atoms-2-productsprogram, which is what i run.
and i'll talk alittle bit about it. we're at kind of a delicatepoint in the atoms-2-products program because we're inthe contracting phase, which basically means i can't tellyou anything about the program. right? i can tell you aboutthe program before we get into thecontracting phase and i can tell you aboutthe program after we get into the contracting phase.
but right now, we're in sortof an indelicate moment. the performers can tellyou, but i can't tell you. but i can give you the generalphilosophy of the program. and the generalphilosophy really is that we've got lotsof terrific technologies that allow us to assemblethings on the nanoscale. and a lot of that's bornfrom huge, huge investments made over the last 15 years. and then we have a lotof assembly technologies
on the right endof this scale that are our conventionalmanufacturing technologies that we're allvery familiar with. but really, there's kindof a giant gap in between. and the atoms-2-products programis really about closing gap, is being able to build thingsall across the length scales, from the nanoscale, theatomic scale, all the way up to the human scale, beingable to build things at the human scalethat take advantage
of nanoscale properties. so when this program gotbriefed to the darpa director, i used this chart a lot. because this is the chart thatshe ripped out of the chart deck and held in frontof her the entire-- i don't think she waseven listening to me. she held this chartin front of her and looked at itthe entire time. all right?
and she was focusing on whatthe strategic surprises might be if somebody was ableto successfully cross that bridge in aneconomically practical way and start building products allthe way across the spectrum. and one of thethings that lead us to believe theremay be surprises in this sort oftechnology is the work we've been doing on somebiomimetic materials work, which is the gecko--generating synthetic gecko
material. gecko's setae arereally interesting, because they are micronscale, right-- in fact, submicron in scale. and they're basicallyjust small-- i call them broccoli--they're small hairs that exist on the ends ofthe feet of gecko. about two million of themper gecko and about 10 to the eighth contactpoints between those hairs
and the wall. and the interestingthing about this is that it's a reusable,self-cleaning attachment mechanism. and it actuallyends up that it's using van der waals forces asthe way to attach to the wall. it turns out it'sreally, really hard to build that structureusing any assembly technology we have today.
so this is what wehave defaulted to. you had that beautifulbiologically assembled structure, that pieceof broccoli, or hair, or whatever you want to call it. and this is actuallywhat we can build. this is what we can build today. and this is reallydifficult building, building these little microwedges. they're about 100 microns long,about 50 microns at the base.
and they taper down to kind of aragged edge that, in some ways, mimics what's on theend of the gecko setae. and the way it attaches, thesemicrographs on the right show, actually, the onlything that's happening with this gecko material,with the flexible microwedges, is they're bending overand touching the wall. so it is indeed vander waals forces that are holding thismaterial to the wall. and if you canget this contact--
and it's actually a bigif-- you can actually get substantialattachment forces generated using this material. one of the messages thatwe got from the-- this was the z-man program--was that scale-up-- and i think this is goingto be an issue for anything that you build at thenanoscale-- turns out to be really, really difficult.we were able to make very, very small patches of this materialthat would be able to adhere,
really, as well as thegecko toes themselves. but as you started scaling up,it went to two by two tiles and then arrays oftwo by two tiles. we ended up having to havesurfaces that were very, very flat to accommodate the flatnessrequirements of the devices. but it did work. but that scale-upcontinues to be a challenge for this technology. and here's the outcome of this.
this is actually--this is the first time we climbed with the paddles. so this is nothingbut gecko material. you'll see himplace it on there, and load it, andstart climbing up. but this was thefirst time anybody had ever climbed on anythingusing-- human-scale climbing-- using this technology. so again, the messageof this is not
to run out and generategecko material proposals. don't do that. that's not the point. the point is this kindof project illustrated to us that there is potentialsurprise in this technology area. because to makethis practical, you need an economic way ofbuilding these very, very small structures, thesesubmicron scale structures,
into a system that can beused at the human scale. and by the way,that's still an issue. we still don't havean economical way of building these. we can build them,but not so that we could make 1,000 ofthem or 10,000 of them. so back to this, back tothe atoms-2-product program. so if there's goingto be a surprise it's going to happenrelatively quickly.
that's what asurprise is, right? and so theatoms-2-products program has really landed on twoapproaches-- or one approach involving two differenttechnologies-- to really see if we can cross that gap. so we have a lot of technologiesin the upper left-hand corner there that are really analogousto electronics manufacturing technologies. and then you have allthese things on the right
and on the upper right whichare robotics technologies. and so what we're trying to doin the atoms-2-products program is see if we can--the upper left here talks about stretchingfrom the small end up. we're really takingthings that look-- these microscale technologiesand trying to build those up so we can buildbigger and bigger. and this is about speed. it's about takingthese technologies
and making them go fasterand faster and faster and building three dimensional--you can call them materials. you can call them devices. i think it's very blurrywhat they actually are-- but building thoseup to the micron scale. but it really is abouttaking technologies that are analogous toelectronics manufacturing and pushing those to largersizes, and at the same time, taking robotics technologyand pushing that down in size.
the smallest thingi think we regularly assemble with ourrobotics technology right now is that there's a 60-microncapacitor, chip shooter, that's commercially available. this is a little moviefrom sri over here. some of these robots areabout a millimeter square. and we're looking at usingthose to assemble the products from the technical area. one performer's onthe top left there.
and that's reallywhat we're looking at to see if there is arelatively quick way to cross that assembly gap. so i'll say one more thing. so darpa hasrecently instituted-- one of the problemswith darpa is that the way toget their attention is to write a proposal. and proposals arean enormous effort.
we all know that. we have a new process calledthe executive summary process. and all we ask you todo is write one page and describe your idea. you can submit itsecurely under the baa. and you'll get a,nah, not right now or you'll get a, call backand let's talk about it. but if you'll invest threehours to write a nice page description of whatyou're interested in,
we'll invest roughlythe same amount of time to take a look atit and give you an indication, prettyquickly, about whether or not we might be interested. so i highly recommendtaking advantage of the executivesummary process. if you want really,really detailed feedback you have to go theproposal route. but if you just want tofloat something through,
please take advantage of theexecutive summary process. so thank you. and i would like to welcome ournext speaker, heinrich jaeger, a professor of physics atthe university of chicago. this is a fantasticmeeting here. and i'm just thrilledto be part of it. so what i want todo today is tell you about a very humble, very,very simple type of material. we call it a granular material.
and in its verysimplest form, you can think of this asjust a bunch of spheres packed randomly. now the key point is thatin this aggregate type of material, the interactionhappens just when the two spheres touch. and so a lot of different stufffalls in the same category. it doesn't have to be spheres. it could be coffee, itcould be sand, coal.
essentially, after water, thenext most used type of material is probably granular material. in all sorts of shapes andforms, we use it every day. very, verytechnologically important. but this is the prototype. now this is asession on assembly, so i should start witha little bit of history. as far as i can tell, the veryfirst publication on this topic dates back to 1611.
so this is johannes kepler. and he's thinking abouta macroscale object. in fact, this is a snowflake. and there's somethingabout this object on the macro scale, thesymmetry, the six-sidedness, that fascinates him. and he surmisesthat there has to be some ingredient, somebuilding block, some essence of snowflake somewhereinside that generates
this on the macro scale. and so he is a goodphysicist, right? he's a reductionist. he plays around and he looksat various arrangements. and he then claims that,for example, arrangement b-- right here, youshove things together like that, with sixnearest neighbors-- then gives you the type of symmetryyou see in the macro scale. this is not meant to exactlyreproduce a snowflake.
but it gives you the keyfeature of a snowflake as seen from this perspective, right? now this is the beginningof crystallography. since then, we use thesame kind of ideas. from the bottom up, welook at local symmetries, local interactions. and we imagine that that willscale up to the large scale. we are very successful withthat on the atomic scale. we can make crystalsof all sorts and kinds.
i'm showing you here, inthe middle, an arrangement of gold atoms where theindividual little dots-- this is a high resolutiontransmission electron micrograph-- areindividual atoms. perfect crystal. now this is then arrangedinto a larger scale object. this oblong thingis a nanoparticle. it's about 10, 15nanometers across. and then, likerussian dolls, you
can arrange thosenanoparticles, or maybe their spherical cousinsthat are the background, into two lattices. again, we're reallygood at that. but these are ordered lattices. they are not manufacturedby little robots right now. this is brownian motionself-assembly that does that. the question is, whatif something goes wrong. what if there is a defect?
what will happen? typically, these crystals, infact, they cleave very easily. they can be very strong,but they are also susceptible to damage. now there's a whole bunchof different materials that are not ordered. and the question thenbecomes, is everything lost. are there stilldesirable properties? and the answerdepends, of course,
on what you're asking for. but there are classes ofmaterials, high tech materials, nowadays used allover-- and we heard about this-- so, glassyalloys, metallic glasses, polymer materials-- thatdo just what we want. they have a high elastic limit. they maybe even have a highstrength or combine both. yet they are amorphous. there is no precisestructural local arrangement.
and that's kind ofinteresting, actually. so you can scale that up. you can ask, what else can wedo when we just essentially pour things together,shove them together. is everything lost? how do we getdesirable properties? well, here's anotherexample on the macro scale. the softness comes aboutnot because everything was perfectly put together--i mean, you could do that too.
and you can get thesoftest spring-- paul showed that yesterday. it's beautiful, right? but in this casespeed, or the inability of the bird to do it preciselydoesn't lose anything. you still have abeautifully soft structure. let me scale up alittle bit more. we heard architects talk, right? so you can be very precise.
you can makebeautiful structures. so this is gramaziokohler's lab in zurich. and they had thiswonderful wall built for the biennale in venice. now you need precision for that. and they use robots toachieve that precision. maybe you ask, whatabout another approach. what if we don't preciselyplace these bricks? and this is something we mighthear later, from achim, about.
so here we also userobots, or they use robots. but they just simply poura kind of building block. it's not a brick in this case. and then you get structuresthat you can actually walk into. you can still makea vault or a dome. it holds up. it holds at leastits own weight. and now maybe the question is,why would you want that, yeah? you can have rapidlydeployable materials
you can just pour into place. you maybe get, also,some aesthetic component from being underneath thisinteresting way that light gets filtered, like alittle bit under a tree. so disorder has advantageson many, many levels. well, the way i like to dothink about materials currently goes along down this axis. so let me walk you through. there are many ways,of course, to decide
how to pick a materialfor a particular purpose. the one that i like is to lookat how a material responds to when you push it, deform it. and there are two modes that arestandard, are really important, to think about. one is you compressit from all sides. and the resistance tothat is this number b, the bach modulus. high b means it'shighly resistant.
you could alsoshear it like this. and the resistance to thatshear is this number g, the shear modulus. now liquids, for example,are hard to compress, easy to shear. so the live all theway on the right. and then other materials, forexample, cellular networks, they are comparatively easyto compress and hard to shear. and you see that there'sa relationship, for most
materials, betweenthese two numbers. they're not totally independent. and so they live in these nichesalong this wide, wide range of this particularratio of b over g. these numbers, they are justindicators of a rough place where they are. but you can't shiftthem around arbitrarily. or can you? so that's the questioni want to ask.
so are there materials thatyou can actually shift across or that can change, thatcan tune themselves, all the way across? and in turns out the simple,humble material i told you at the very, very beginning,this granular material, is just one of those. so for example, you canhave it in a configuration where we call it unjammed. parties can movepast each other.
it almost flows like a liquid. and you can jam it together,push it a little bit closer. and then it becomesrich like a solid. so this is what we calla jamming transition. it's purely geometry. can particles get pasteach other-- yes or no? it's like a traffic jam,almost, for materials now. and you can go back and forth. and the interestingthing is that there
is no structural changeassociated, in this case, from going from aliquid to a solid. it is amorphous on both sides. it doesn't turn into a crystal. it doesn't need to. and that's very,very interesting. now we're allfamiliar with that. so you go to a store, you buya pack of vacuum packed coffee. it's hard as a brick, right?
you can hurt somebody. and then all you do, you snipopen a corner, it flows out. you didn't changethe temperature. you didn't do a thermodynamicphase transition. all you did is you alloweda little bit more space. and particles can flowpast each other unjammed. now why is this all happening? well, in the interior,there are these contacts between particles.
and they set upthis beautiful vein like network of stresspaths or force chains. so this is a computersimulation from a few years ago. i'm putting someload on the top. and i'm asking, how are thestresses transmitted through. and the dark lines are wherelots of contact stress is. and the light lines arewhere there are fewer. and you see that thereis inherent disorder. it's inherentlydisordered, amorphous.
and the very interesting thing--it's just like in a cathedral-- you have keystones and youhave key particles here. if you zoom in, let'ssay, on the right here-- i'm zooming out. see where my cursoris-- this particle here essentially experiencesvery few forces because there issome other particle above that shields the loadand puts it to the side. now this is a fingerprint ofone particular load application.
if you change yourload from the outside it will have another pattern. it responds dynamically towhat happens on the outside. and that means itcan be self-healing and it can be adaptive. but it's a very, verysimple material, actually. so let me show youa little movie here. so we whack atwo-dimensional assembly of this-- this is nearlycrystalline, actually--
with a hammer. eh, film it at 65,000frames a second. and you see how, in thiscross-polarizer image, the white assistedhigh forces, how they come up and then disappear. it's ready for the next blow. so this disorder is actuallyreally helpful here. there's no cleaving, noeasy or immediate cleaving. it essentially has micro-cracksbuilt all over it--
just what you want. so what can you do with it? so let me giveyou some examples. here is one idea thatcame out of-- well, we have to say it--another wonderful darpa program, their programmablematter thing from a few years ago. the question is the following. let's suppose you wantto secure and pick up
an unfamiliar object. now of course, if youknow the object already, you can build a perfectgripper for that. let's suppose youdon't know the object. is it brittle? is it soft? is it hard. you need many degreesof freedom typically. you need to havea lot of feedback.
and you need some brainor some big computer somewhere to steer all that. there are beautifulsolutions for that, but they arecomplex, complicated. can this jammingtransition possibly provide an alternative approach? and the answer is yes. so here we got abag, elastic bag, filled it with thisgranular material,
and then we evacuated. and it turns intothis instant mold. it memorizes, essentially,any particular application of load from the outside. so what we're doing,we're essentially confining the volume withcrowding these particles in a little bit so they jam. that what this is. and then when you let theair back in, it unjams.
so it's one way of introducinga jamming transition. now what that means is thatif we now take such a bag and run it over anarbitrarily shaped object, it will immediately conform. and then withoutreprogramming anything-- an open loop-- you can runit over some other object, for example, a brittle object. notice, by theway, you don't even have to run it around that.
there's some suctionthat is a little benefit of this approach. and this was really a raw egg. so i'll show you that. so this is somethingthat got spearheaded by my then postdocbrown in collaboration with john amend andhod lipson at cornell. now in fact, johnpushed this further. and there's alittle company now,
a start-up calledempire robotics. and they sell that tothe automotive industry [? and not us ?]. so in the old initialversion, v1, on the left, you evacuate it a little bit. and then you let theair back in it unjammed. now you can let the airin-- we can actually pressurize it a little bit. and that allows you to play thisbeer pong thing on the right.
so now, rather thanthrowing an object after you've picked it upby using your elbow here, you just basicallypop it out like that. and remember, we have arandom amorphous material with all sorts offorce chains in there. but statistically,right, since we know how to dealwith jamming, it turns out we can actuallyget this to work. on this level, it has precision.
so let me go another stepfurther here, with that. in collaboration withour friends at irobot, we made the world'sfirst totally soft robot. so you take your jamming cells,but around a linear actuator with that. so it's one degreeof freedom this way. but with adding jamming, we turnthe bending modes into action there. and what you can dowith that is shown here.
these legs areindividually controlled. this thing is still tethered. it can run. and this is really, basically,a robot made out of sand, except for the skeleton. so we are familiar withgranular materials, aggregate matter of thistype, on all sorts of scales. so on the upper left, thesethings are 20 tons each. and there's a reasonyou want them jammed.
same with the packing peanuts. and then, of course, nowadays,for 3-d metal printing, again, it's granular material. that's reallyimportant, how it packs. and maybe at somepoint, you want to ask yourself thefollowing question. or let's just start this way. anything will jam if i justpush it together enough and if it's hard enough, right?
if it's too soft, it won't. but what if i don't justwant to have jamming per se, i want to know whenit jams, how it jams, exactly how much resistanceit will have once it's jammed? then i do actually needto think about design. and that's the lastpart of the talk. so this is a real challengenow, because we're not dealing with crystals. we're not dealing with thestuff that we know about.
we're dealing withinherently disordered stuff. but if we could dealwith it, then we would have a reward, right? we would get access toall these good things. and let me just bringthat out one level higher. maybe this is alsoin response to what i heard yesterday a little bit. so we are really thinking,in materials science, about this connection.
maybe it's a pyramid. but there are these two partsof the structural property relationship. we believe thereis such a thing. i wouldn't do science otherwise. and what we really likeis a prescription that goes both ways, back and forth. it's hard, actually, to gofrom the right to the left. it's usually we gofrom the other side.
and especially for theseinherently disordered materials, it is almostimpossible, right now, to go this way, to go fromdesired properties to what i should put in there. so what should i putin if it's not spheres? what shape should things haveto get a particular property? we actually don't havea good prescription. but in some sense,you might say, this is the multi-dimensionaloptimization problem.
and there may be waysof dealing with that. and one way we justheard, also, is evolution. biology kind of does that. so why don't weevolve materials now? actually, we shouldtalk about that later. i'm not so sure that youcan yet do it in real time. but we have some ideas. but we do it in a virtualenvironment on the computer. so it goes back andforth in a loop.
we have a populationof possibilities. and then we're sifting them outin a pretty darwinian process. this is work by mystudent mark miskin. and so when you're done,you print these things and make sure they work. we start simple, withshapes that we build up like a chemist maybe woulddo, but now on a macro scale. and then we can askfor design goals. and let me just give you oneexample and i'll finish up.
so here's the task. it turns out, in this business,it's a very hard task. find the shape of a molecule--it could be any shape-- that packs the highest densityunder a very, very, very simple operation. just pour it in a box. it's actually not known--well, it wasn't known. so we let the computer loose. and here's what it did.
so it moved around, to start,with aggregates of 10 spheres, whatever. pretty soon-- thisis, you know, it gives you the worst, themedian, and the best-- it converges to this particularmickey mouse like shape. who would have known that? so that's the shape thatpacks, actually, best under a particularprocessing condition for a particular goal.
and then once we have that,we can actually find out why. so what actuallyhappens is the algorithm discovered-- itwasn't programmed-- and it discoveredcertain correlations in disordered packings. it's a bonus. you get science out if it too.you discover something new. but anyway, i'm out of time. and i'll thank you.
the last speakerfor this session will be lodovica illari. lodovica illari is ameterologist and faculty at mit and runs the synoptic lab. thank you very much. as i say, i'm not amaterials scientist. i'm feeling i've learneda lot in these two days. and i feel a little bit nota specialist in this field. but i'm going to talk about someexperiments which address some
of the themes of this meeting. first of all, we've beentalking transformation. secondly, we've beentalking about assembly and also transferring, as sklyarhas mentioned in his talk, trying to make people understandwhat is not intuitive. and i want to start here. i teach meteorology. and then my real specialityis large scale dynamics. i'm very much interestedin big storms.
the past winter was good for me. i loved it. i'm italian, but ilove cold weather. but anyway, whati'm showing here is a satellite loop that you'reprobably very familiar with and have seen many times. i don't want to describe it. just to give you an idea, we aregoing from the very small scale that people have been talkingabout in biology to the earth
scale. which, embedded inthat earth scale is embedded a differentscale of motion. and i'm going to talk aboutsome of this regime of weather that you observe on earth. and i'm talking about somethingthat i use in my teaching. we call it weather in a tank. it is a teachingof rotating fluid, because we are ona rotating planet,
using very simpleoperators, which are a turntable with aco-rotating camera on top. so when we are onearth, we are standing. and we are rotatingwith the system. and when you try toteach the students how the system behavesin a rotating frame, the math is quite complex. this is called geophysicalfluid dynamics, gfd. it attracts a lotof mathematicians.
and when you went to explainit to, say, undergrads, you don't want to go intothe complexity of the math. but you want to workthe physical idea. so we have developed-- togetherwith professor john marshall, we have developed thisapproach, which is experimental, to try to understand someof the weather regime. by the way, i alsowant to mention, some of the visualizationsthat i'm going to show are from bill mckenna, whoactually took one of my classes
in his senior year. but he's an architect from mit. so we are very much interestedin the effect of rotation and visualizing it. so to go next, what is thematerial transformation? what does rotationdo to a fluid? and to show this simply, i'mcontrasting two experiments. on the left, you havea non-rotating tank. this is a tank of about 60,70 meters standing on a table.
you have the sideview with the mirror. and to see something there, wejust agitate it with the hand. and then we addsome dye droplets. on the right, instead,the similar tank is on a rotating platformsimilar to the one i've just shown with a view fromthe co-rotating camera. so on the top, you see imention the co-rotating view. so what you're seeing isin the rotating frame. and the side view is alsoin the rotating frame.
now the tank isbeing spun to what we call solid body rotation. so you have to spin it forat least 15 or 20 minutes to get all the fluid goingaround together with the tank. and i'm just going to show you,very simply, this experiment. and i'm going to pauseit in between just to notice the difference,to contract the non-rotating with the rotating. what you expect-- after youagitate, you put in the dye.
and if it'snon-rotating, the dye will try to spreadin all directions. it does not happen inthe rotating fluid. if i carry on, you notice youget this filament, this laminar structure. look at the differencebetween the two fluids. so the only thedifference is rotation. there's no other differencebetween the two experiments. and if i keep it going, i wantto also show you the structure.
the filaments interminglein the various fields in between themselves. so the fluid is very different. and it doesn't actually-- thisis a better, higher resolution image from the same experiment. what you notice isthe filament nature and what we call aribbon, kind of a curtain. sometimes they're called ataylor column or actually a curtain, a [inaudible].
so the fluid is stiffened inone direction by rotation. so instead of spreadingin three dimensions, it's actually movingin two dimensions. and it's behaving morelike a solid than a fluid. so it's not really a fluid. so when you're talking aboutgeophysical fluid dynamics, the effect of rotationis very important and makes the fluidlook different from a normal standard fluid.
so the connection here-- look. this is the same image fromthe experiment i've just shown. and that is jupiter's red spot. jupiter is fastrotating, very much influenced by the rotation. what you see isstrikingly similar. so i did the control experimentwith this very simple setup. and i'm getting patternsthat resemble jupiter. it's not a simulation, clearly.
it's just there to show youthe importance of the rotation on the behavior. so then i'm movingand saying, ok, we've been talking about tryingto do some assembly. and so in this control frameof this tank experiment, i can play with ingredientsand make some laboratory abstraction of weather. so my ingredientsare very, in a sense, dictated by the large scaleof the atmospheric system.
and we know that the drivingis the temperature difference. so the pole is very cold. the equator is warm. that is what drives pressuredifference on the atmosphere. so the motion is dictated bythe temperature difference. but we are on a rotating earth. so we have to thinkabout that fact. so i'm going to dotwp experiments. and i'm going to startfrom the omega is small,
rotation's small, delta t large. so i'm only changingthe rotation. i'm going from small to largeand keeping the temperature difference the same. this omega smallrepresents the tropics. because if you notice, the axisof rotation is at the pole. and what matters is the rotationin the direction of gravity. the fluid is stratified. and it's the rotation inthe direction of gravity
that the fluid feels. so at the pole, you feelthe maximum of the rotation, because gravity and the axisof rotation are parallel. if you go to the equator,the gravity and the axis of rotation are perpendicular. so there is hardly-- no, itdoes not feel the rotation. there is no rotational effect. as soon as you moveaway from the equator you start feeling the rotation.
so if you, for example,go to 30 degrees north, you will have aneffect of rotation. but it'll be small. if i go at 50 degrees north,like 45, 50, where boston is, you start feelingmore of the rotation. so there are twotypical regimes that we can play with, the deltat large and omega small for tropicalcirculation-- we call it hadley cell circulationin the jargon--
and the omega large, which isthe middle latitude weather system, say, theboston weather system. you could think about those. so the first example isfor the small rotation. and the delta t is acquiredby putting ice in the middle. i'm not warming the outside. i'm keeping it a constanttemperature-- a room temperature-- but i'mcooling the middle with ice. and then i let it evolvein a rotating frame.
so what you'reseeing here is still the view from the rotatingframe from the top and the view, in the rotating frame,from a camera on the side. the movie's been sped up. but in the actualreal time, in the time to cover about 60centimeters it's going around for about 60seconds per revolution. so it's moving veryslowly, very slowly. so if i let thisgo, i want to notice
that-- i put pepperdots, by the way, on the top to show the movement. and i put a bit of green dye. so the dye has got some density. it falls on the back. it falls down a little bit. but it starts spiraling. do you see the spiral movement? and then look at this videoand see the pepper dots.
the ones close to theeyes are moving faster than the one outside, showingthat there are strong currents. at the top. and the fluid is spiraling down. i've also introduced somepermanganate crystals that will go to the bottom andgive me the flow at the bottom. so the flow at the top isfrom the pepper dots moving. the flow at the bottomis given by the trail, or the permanganatewhich goes at the bottom.
so if you let it go, i cansee all the strong currents and at the bottom, younotice the trailing is in the opposite direction. so there is what we call shear. the flow is verystrong at the top and weaker, because offriction, at the bottom, but also in the other direction. so the fluid inside adjuststo the spiral shape. and in my next slide,i'm summarizing
the beginning, the spiralingaround, and the final state. and so you see the dye is goingaround because of conservation of angular momentum. because what ishappening is the fluid is sinking close to the cold,going out of the bottom, coming up, and coming back in. so when the fluidat the top comes in, it goes close to theaxis of rotation. it's like a skaterbringing their arms in.
it spins faster. the fluid at the bottom goesaway from the axis of rotation, spins less. so you create a corkscrewshape into the fluid. now the analogy isstriking with you. i don't know. i'm a meteorologist. upper level winds arewesterly in meteorology. they blow from the west.
they blow from west toeast in the same sense of the rotation of the earth. and they are the upper levelmovement that you notice here. the surface windstend to be easterly. and they're calledthe trade winds. those as the windsthat were used to trade from europe to america. so you move your ship followingthe easterly flow to go. and when you come back,you try to go a little bit
north to pick up thewesterly flow, some of the westerly flow. so the trade winds areeasterly at the surface. and the top ones are thewesterly, the jet stream. and this is just conservation ofangular momentum in a rotating system. now i'm moving on. and i say let'smove towards boston. let's go towardsthe middle latitude.
and so my omega is large. my delta t is still the same. so what i've got hereis the same experiment, but i'm going to stopit after i put the dye. because immediately, youknow that the previous fluid was very laminarwith axis symmetry. this fluid is notbehaving in the same way. so the only difference isi increased the notation. i didn't do anything, justincreased the rotation.
if i let it go, you startseeing the colder going to the outside. the green goes out. the red comes in. so it tries to equilibriate thenorth-south-- the latitudinal-- temperature difference. but it does it in akind of chaotic manner. so you get interminglingof cold and warm. and i'm stopping it here toshow those cold and warm will be
the front coming into boston. for example, here,imagine boston is here. one day it's warm andanother day it's cold. this is because ofthe weather system. these are the weather systems. they come about in ourlatitude because we are feeling the earth's rotation. so in a sense, the fluid isnot axi-symmetric anymore. it's not laminar.
it's becoming turbulent. and the transitionoccurs because we increased the rotation. so it's two completelydifferent regimes. so a big transformation by justincreasing the rotation rate. and i wanted to finishit to show you, then, the connection withthe atmospheric data. so if i let it go,you see the evolution. the fluid tries to mix inthis organized kind of manner
with-- i could call them-- thesehigh and low pressures that are alternating around the pole. so if i summarize this,the air circulation is no longer axi-symmetric,but turbulent. there, these aretransporting cold fluid away from the cold center towardsthe edge where it's warm. and vice versa, thewarm fluid comes in. so if you notice, iput the red outside. and now it's close to thepole in the final stage.
and the green that was closeto the eye has come outside. so how does it compareto a weather system? this is a temperature fieldon the planetary scale. if you actually lookin it carefully, the north pole is here. this is north america. so it's the polar viewof the temperature field for a week in winter. it's not this past winter,but a similar winter.
and i want to show, for example,here, if i stop my-- oh, i'm sorry. i did a mistake. the mouse has disappeared. where has it gone? yeah. so if i-- sorry. no. so i cannot go back to that.
doesn't matter. so the analogy isthat the temperature field-- no, i have to go back. otherwise it doesn't make--i'm not teaching you properly. so why is this doing this to me? so the warm air is comingnorth, if you notice. and the cold air is coming downin a swirling motion, very much similar to the one thatwe've noticed there. so in a sense, there isa hint of circulation
similar to the one. and you might be more familiarwith seeing the weather system in a satellite imaged map. here is water vapor. so what is whiteis warm and moist. and what is darkis cold and dry. and so if i stop this moviehere at this time, for example, and let it go again,i can see that there is a storm coming along.
so where my mouse is, i havewarm, moist air coming along over the south of the us,and then cold, dry air behind with a cold front. and i can count some of these. if you count them, you cancount several of these weather systems around the globe. they generally go between six toeight of these weather systems around the globe every time. and you can see the evolutiongoing from west to east,
moving with the jet. so to make the visualization,the experiment, available not only topeople who have the tank, we have made avirtual laboratory. so we have reconstructed theexperiment in a digital form by imagery, taking thetop view and the side view and reconstructing it. and this is the workof bill mckenna. i'm going to finish soon.
and it's showingthe visualization of the plume in thefirst experiment where we are reconstructingthe spiral shape of the plume. and in the second one, we arereconstructing the revolution of the eddies coming out. and these use the fantasticsoftware from-- he's well known from hiswork in architecture. and we find thisvery interesting. and we are actuallyconsidering this,
to put this as apossibility of making a connection withthe tank experiment and making it available topeople that don't have the tank but want to see thevisualization in this form. so i just wanted tobring it back to reality. normally, in atextbook, you will have this imagewith two regimes, hadley circulation, eddiesin the middle latitude. and the students connot--it's very difficult for them
to understand the transition,why these are different. when you do the math,it is not so simple. and the experiment makesit much more alive. so in the theme ofthe meeting, i just wanted to say that thisis my transformation. i've learned that there ismuch more complex information, much more. but i'm learned a lot about it. and so the question then,for you-- and i might ask it
to the panelafterwards-- supposing i want to do a 3d printing ofmy experiment, if i go here and i've got thissimulation and i want to do some 3d printing--or instead of 3d printing, can i put somematerial in my tank and come out with a shapethat looks like a spiral? thank you, lodovica. we're now going to havethe panel discussion. so we'd like to call all thespeakers to sit on the panel.
and then we'll have-- takealso a short time for q&a. so we're running late. so i said we're going to shirtthe break 15 minutes later. so 11:45, we'll have the break. and it seems thatlodovica already has a question to start. so i'll leave it to her. so i learned a lot. and this is clearlyis not my field.
but if you are justlooking at that what you're doing from ateaching point of view-- for example, skylarhas mentioned that he was putting neutrallybuoyant particles in his fluid. and we've been doingthis in this experiment. so you can embed all the-- intothe fluid neutrally buoyant and look at the shape. you can look witha laser and try to measure thevelocity of the fluid.
but this is research. but if i want, instead, to usesome of this fancy material that you have andi want to create a shape-- becausestudents learn much more from visuals than fromequations, and from practical. but not many will havethis tank experiment. so we could createsome shape and make them related to-- say,for example, in this case, it will be the weather regime.
but there will be otherfields in physics or biology. well, let's have a trade. so i'll help you on thematerials and the printing and you have to help mevisualize the dynamics of the fluid in our tanks. because that's one ofthe hardest things. like, if we change thelocation of the pumps, you can't see what's happening. you can't.
because you have a turbulentfluid in all directions. what i've done isi've organized it. so what the rotationdoes is actually have an organizing principle. it makes easier. it's different. so it's organized. although it lookschaotic in a sense, it's a kind of organizedchaos in my case.
i love the idea of, whatcan we do to your tank to have a smiley face show up. or what can you do to yourtank to have some shape, whether it'spredetermined or not, but emerge showingyou really have control of thedynamics of the system even though it's so complex? sure. you don't want to disturb it.
so you need to have a materialthat is [inaudible], that is probably like you said. it goes in. and then it adjusts to theflow without interfering. i'm curious about-- in asense, it's an analog computer that you've developed. and i'm wondering howfar you can carry it. so for instance, thetopography of the landscape is not part of themodel currently.
but you could, using 3dprinting, build a model that-- the surface and the ridges. we actually canput the topography. or you can actually putan obstacle in there. you can actuallymimic the difference between a north-south differencein the rotation with a sloping surface and then put the bumps. we call it flow open obstacle. so you will have-- like,imagine the rocky mountains.
the rocky mountains are abig obstacle in the flow. and they span from polereally, high latitude, to low. so you can havea sloping bottom. and on the slopingbottom, in the tank, a sloping bottomgives you the effect of the rotationchanging with latitude. and then on the topof the sloping bottom, you put a ridge, a little ridge. and the students look at wavesthat propagate on the flow.
so we're using that. so there's anotheraspect, i think, that came out ofthis first question. so i guess, one, yeah, itprobably can be printed. and then you getexactly what you have. the inverse, i think,is very interesting too. what if you want to use thisvery complex hydrodynamic flow to assemble something? and so that goes to a questioni have for you, arthur.
thermodynamics worksbeautifully, brownian motion, like you said. but it's essentiallya uniform heat bath. that will give youassembly, as we know. we're all assembled that way. but this adds thisfar from equilibrium, non-equilibrium aspect. so you drive the system intocertain corners of phase space. and i think thebig question for me
is, can be utilize thatto speed up things. usually, turbulence isused for effective mixing. but that's maybenot what we want. we want effective assembly. can we use it that way? so two things aboutthat that i thiink are interesting-- one of themis that that last thing that i showed you wasn'tturbulence, but it gave-- it had a turbulentimpact in the sense
that it was chaoticbut organized motion. so it wasn't necessarilyhydrodynamic. the other thing, i think, ispeople are starting to realize. i mean, a lot ofpeople study cells plated onto a glass slide. but now they're startingto be able to get more three dimensional views. and what they're finding is thatthese motors and these actin networks and so forth areactually almost breathing.
so there is some push, somehydrodynamic push in cells. and what the impact of thatis on the self-assembly, nobody really understandsat this point. it reminds me, in the tank,we've been thinking about, if we created a certainpattern-- in the big tank it's a crystal. and it's accumulation. it's growing, growing,growing, growing. one thing we can'tcontrol at all
is the boundarycondition, at the moment. like, we couldpotentially control the lattice and how pureit is and that stuff. but it's really challengingto make some arbitrary shape that you wanted to use it as aproduction mechanism assembly. but potentially, with theturbulence, you could do that. by understanding thecomplexity of how it's flowing, you might be able to geta structure assembled that has arbitrary complexitybut also that lattice assembly.
so the chair that you'remaking, it assembled because of the shape of the--so how do you create the motion of those-- basically, theturbulence is, let's say, chaotic at that point. we're not really controllingthe pattern so much. the shape is important. so they can't connectto one that doesn't fit. and the pattern of attraction--so the polarity, in this case,
or in the velcro, if itsticks or doesn't stick. so by having the right shapeand the right pole, then it will connect. but it does connect,sometimes, in the wrong way. and art really showed thisabout error correction, that it's going to breakoff because it's weak. and only the strong willsurvive, essentially. and it'll grow strongerand stronger and stronger. but it also reminds meof the sound studies
where you play adifferent frequency and different soundstructures emerge. so maybe we can use thatturbulent environment to produce usefularbitrarily complex things. so i wanted to make a commentabout the educational aspect. because i think it'san important thing. and in fact, it's whatdrove me into 3d printing. because usually, you make3d prints to do something or to test something.
but you can alsouse it to visualize. and you can make multi-componentmodels like the virus-- and i've got apeptide folding model and so forth-- that we use toteach our graduate students. and they learn something--they've already had biochemistry. and molecular biologyis undergraduates. but none of themhave learned it in this structural, physical way.
and it's a totally differentway of actually getting stuff into your mind. and there's this, iguess, recent trend on the idea of perceptuallearning, in other words, that we don't justlearn in one way. we learn with our bodiesin ways that none of us really is cognizant of. but it happens. the physical isso important, i've
found, as well as thevisual, especially in these non-intuitive things--to be able to tweak the dials, and play, and watchthings emerge and go back. play with parameters. you learn a lot. i was thinking about theturbulent environment. john, for yours, onecritique would be, like, in the components movingaround to produce devices, it's still a top-downassembly method in my mind.
because we'redirectly controlling every location of every piecein order to come together in a larger assembly. there's nothing wrong with that. i'm sure it's going to befaster than the traditional way. but the conceptual challengeis, can we lose control to gain. can you lose a little bitof control and in the end have properties arrive thatyou could have never gotten to by having direct control?
and that's always, in the backof my mind, the holy grail that you can get to. and i feel like thedirected assembly way, it may not get there. so i don't think we knowthe answer to which way is going to work out better. the key element inatoms-2-products, it just turns out to be speed. and right now ithink if something
like directed assemblyis going to get used in that kindof application, we've got to crackthat speed variable. if there were a credibleway to go after that, i think it would be of interest. [inaudible] maybe you go first. and if there's questions,raise your hand. yeah, ok.
and after that, we go to them. well, i was about to say thatif you want to ask questions-- so anyone, raise your handif you have questions. but you had a question. but i think oneof the key issues though is, if we thinkabout self-assembly, the yield is nevergoing to be 100%. that's something we'llhave to live with. but i think there'sways to increase speed,
like having more parts inthe soup then you need. by having the right container,the right environment, understanding thedynamics of the system, you can get higher yieldsby weeding out successes. so don't let yoursuccesses block. there's all these little tricksto get much, much, much faster. and because you'remassively parallel, maybe it's ok to not have 100%. i agree.
well, we live like that. yeah, exactly. well, the yield's not 100%the other way either, so-- we have a question. hi. so i work a lot withrobotic assembly. and one of the thingsthat we see-- and i have a question for all of you. how do we design the interfacebetween the components
that we're using to actuallyassemble these structures? so at chemistry,it's always fields. in biology there's somekind of combination of fields and structure. at macro scalesi'm having trouble. i want to usefields, but i'm not aware of fields that operateat more macro scales. and so john's discussionof the gecko feet is saying go macro and goall the way down to micro,
nanoscale. i'm just curiouswhat ideas you have. or there were couplediscussions of this, of well, maybe jamming doesthe thing, maybe shear rates does the thing. i would argue, maybe you wantthe inverse of the shear rate to actually get astatic structure. but what tools do we haveto identify interfaces between these componentsthat we can have
proper structural properties? i have a thought that's notanswering your question at all, but is related to that. like, one challenge withthe chair, for example, is that you need to have thisreally complex set of joints-- especially if you want to makean arbitrarily complex shape-- you need to have differentjoints at every interface such that only theright once connect. and that becomesan enormous problem
to figure out how manyunique parts you have, how many uniquejoints do you need, how do you make sure they'regeometrically and polar unique. and so we're justscratching the surface. but thinking about a languagein order for this to happen. and you give it a shape. and it propagatesdifferent patterns of geometric andorientation of poles such that it will alwayshave the correct pattern
in each one. and it's just aninterface challenge that you have to figure out. and so we built a chair. it has this very humblenumber of unique connections. and then you cango up from there. but if you want to makeregular repetitive structures, it's much easier, becauseevery one can be the same. but it doesn't answeryour question at all.
this is the one-- any timethat i start a darpa program, there's always thingsi'm very hopeful about. there's always somethingthat scares me. and this is theone that scares me in all of this is, whatabout the interfaces. i've seen mechanicalsnaps down to the micron scale, which are reallyinteresting and inventive. below that things get very--i'll just stick with scary. maybe jamming has thesolution, the gripper.
it really dependson what you want. so if it's ok to have adisordered local structure, or at some scale,disordered, right, let's say, within the grain,i didn't talk about it. the idea is that it's thegrain to grain interface. if that is ok tohave disordered, it gets you speed,serious speed. and if you're happywith the properties, if you want somethingthat's maybe tough
but not the strongestthing-- so you're ok with a non-crystallineobject-- you're fine. but i guess it's a trade-offof various desired outcomes, really. has there been anything that'sswitchable lattice to jam? like regular toirregular on the fly? because you couldthink about, if we could have a shape morphingparticle, you could do that. you could probably induceit by other activators.
but can you go from crystal,back to jam, back to crystal? and then maybe you couldget the best of both worlds. it's a good idea. let's do it. let's do it together. i was just going to add--so i don't know in general. but for example,you're asking, what are the design principles forthe macroscopic fields i'm going to be using toassemble that, right?
so at least in clotting, thispolymer's actually pretty big. and so it is designedexactly at a length scale where it couples wellenough to the flow field, but also still hasa microscopic part. if you want to do this with verytiny objects, it doesn't work. so there temperatureis an important one. and so those are thetype of considerations that you have to deal with. because if not, either your snapthings or the amount of energy
you will have to put in willbe insane to actually do the transformationsthat you're looking for. i mean, at large scaleit's got to be gravity. if you go large enough,the field will be gravity. and it'll work. the work i'm doing isnot stochastic assembly. so we're trying to buildspecific lattice structures. and you can usemechanical snaps. but then you're limitedby the elastic modulus
that's enabling the activationenergy of that snap. so you're not taking advantageof the maximum material properties thatcould be available if you rely on a snap. and in magnetics it's goingto be highly limited as well, right? you only have somuch magnetic force that you can actually storein that small component. and if you're trying to do itwith lightweight structures
or super strong structures,you run into the problems that i'm just throwing out. this seems to be a majorsticking point that is the elephant in theroom of the whole problem. yes. another thought is, like inthe chair or the really big dodecahedron that artand i did, the magnets are never the strength. we don't use the magnets inorder to produce structure.
you only use themagnet-- and you actually don't even use the pullingforce most of the time. the velcro shows that. so you just need to stick ifyou get in the right spot. but you don't have to usethe stickiness for structure. and we've talkedabout this a lot. matt, by the way, is alsoworking on the lattices. and so that's where thetank idea came from. but maybe we canuse the stickiness
to stay when you're right. and you use geometryand materials in order to get the structure. even by tapering, instead ofhaving a connection like this, which will shear, by taperingit in one way or the other, you're not going toget the same shear. so we can use the restof it to be strong, but have a stickinessin another factor. can i ask, alexander,if i put your polymer--
so you put thepolymer like the one that you're describing, which isvery shear sensitive, correct-- so if i put it in my fluid, whatare the experiments i could do with a polymer likethis in a tank, not on the level of a vessel? it will shear thicken rapidlyin areas of high shear. so that'd be good. how expensive is this? well, it depends.
for a tank like that,you'd have to go and get a cow or something like that. i'll do it in a smaller one. i can do a smaller one. right now it's allcoming from blood. all coming from blood. it's not synthetically made. so our time is up. we need to go for break.
and we'll meet. thank you for theamazing discussion. and then we'llmeet, everyone, back at 1:00 for our next session.