CDFAM NYC 2023 · New York · 14–15 June 2023

Computational Mapping of Biomimetic Structures

Listen as a podcast

Abstract

This recording is from the CDFAM Computational Design (+DfAM) Symposium and features Matt Shomper, CEO of Not a Robot.

Presentation Title: Computational Mapping of Biomimetic Structures

This session is especially beneficial for professionals, researchers, and academics in the fields of computational design, additive manufacturing, biomimetics, and medical device design. Those interested in understanding how computational methods can be used to map and replicate bio-inspired structures for medical device innovation will find this presentation particularly informative.

Please note: The CDFAM Computational Design (+DfAM) Symposium is a respected platform for the exchange of ideas and breakthroughs in computational design and additive manufacturing. This recording is made available for educational and professional development purposes.

Transcript

From the speaker’s corrected captions. Each timestamp opens the video at that moment.

Read the full transcript · 3,200 words

0:07 That is the name of the company, and basically what we put inside the human body as being a very artificial thing, versus what the actual body is, in and of itself, and what it needs. So what I’d like to talk to you about is really more of an ex, a design exploration, and less instructional, because I think a lot of the current algorithms, I mean you guys have heard about a lot of software companies doing a lot of really cool things, and this isn’t really intended to be again like a road map to getting to a final solution, but really more of an exploration on how can we learn from the way that biology works and maps itself, and is it applicable to what we’re currently doing, what are the things we can learn.

0:50 So a little bit of an agenda, a little about me, lesson of the trees, I’m obsessed with trees at the moment, and then we’re going to talk about bone, which I’m very familiar with. Both of these are gonna, we’re just going to look at the structure, we’re going to take some time to figure out, like, how does computational design play with the structures, can we mimic them, how do we do that. No questions though, Duann doesn’t want to do any questions. So quick, who am I, what do I do, just computational design, med devices, that’s all you need to know, look at my LinkedIn, I post way too often.

1:21 So I want to talk a little bit about engineered versus biological structures, because we, when we talk about biology and nature, there’s this iterative loop, that’s what I’m going to keep coming back to, this idea that biology is ever changing, it’s evolutionary, it never stays in in stasis because it responds to external feedback conditions. So we looked at engineering structural, you have an initial problem, right, that’s basically what people pay me to do now, right, I’ve got a problem, here’s my bounding conditions, solve this thing for me. So we may do some simulation optimization there, there may be algorithms associated with it, there’s simulation, there’s design parameterization, maybe you’re using machine learning or statistics, things like that. There’s this refinement section where you basically say, all right, I’m gonna, I’m gonna further refine this based upon the results that I get out of my analysis, and then, and you started to do a final validation to confirm, like, yep, I solved your problem, here’s the solution.

2:19 But here’s what I actually think is interesting, is it’s not really a loop, and that the loop is confined to this refinement, because once you produce the final thing in additive, it is a, is a first order static solution. Very rarely, like you heard from the spirit before me, which was an awesome presentation, does, does it biologically respond to whatever its environmental conditions are, it’s static. I mean, we print metal, we print plastic, those things don’t change, they just, they’re stuck that way. And you know, I see so many people that are like getting this inspiration from biology, and then they print in titanium, and it’s like, well cool, you know it looks organic, but the whole reason why bone is bone, and is shaped like bone, is because it’s vascularized, it has blood flowed, it could change, it breaks, it repairs itself.

3:03 So when we look again at a biological structure, it, it exists, it, it’s sole purpose is to exist in an equilibrium state, right, stasis, so that it’s in harmony with its current surroundings. And like a tree that is, you know, blown by the wind, or in storms, it continually adapts to its surroundings, so it’s it’s seeing a force, and it’s adapting to that by saying, hey, I see a force, and I am no longer in harmony anymore, it changes its structure. So its harmony gets disrupted, it evolves its structure. So we talk about like long-term evolution, but there’s an evolutionary process that happens in each of our bodies every second, and that’s just the body’s adaptation to your local surroundings, and then the structure itself sort of enforces and trims, so and forces, in areas where it sees some sort of a loading, and it gets rid of the things that it doesn’t actually need, and then again, feedback, it puts itself back in harmony, and then it’s in a happy spot.

4:02 So what biological systems produce is an nth order dynamic solution capable of change, which again, I’m so very happy that that Natalie was right before me, because she’s investigating these biological feedback, feedback mechanisms, right, and that’s fundamental in my opinion for the future of computational design, as we consider biological printing and things of that nature.

4:26 So let’s talk a little about my current obsession, that’s trees. So this is an old, I’m going back to, you know, 19th century graphics here, you may bend the sapling but not the tree. I I actually started getting obsessed with this because we we started a tradition where the kids every single weekend want to get up and do a fire in the morning. So I’ve got a, my oldest son, he’s almost seven, and he can like start a fire by himself now, you know, I’m just like let him, I’ll get out in the morning, and there’s like a fight, really can’t get him apart, like green branches got some moisture in it, like you can’t bend it. And so I was like, hmm, I’m, you know, I’m not sure, so I started to do a little bit of investigation on like why saplings are, green wood are so incredibly resilient to breaking in general.

5:06 And so again, this is something that’s been used from a cultural significance standpoint, but in the Ozark Mountains there’s a lot of these sort of tree markers for pointing to specific locations or trails, and basically when trees are saplings, they’re shaped to a specific shape, and then as they grow, they retain the characteristics of essentially what their environment is, they’re forced into a box, and they adapt themselves to their surrounding. So the internal structure of them allows for significant strain, well past the point of what would be considered normal in a engineering structural problem, while maintaining the cellular connections, being totally vascularized, and growing that way essentially into a full structure where they finally solidify.

5:52 So again, just like incredibly cool to look at the microstructure, and when we start talking about the softwares now, well, number one, the processes are capable of creating incredible geometry, and now we have software that that’s capable of actually taking advantage of the the geometry that we can create, and create these incredible hierarchical structures, right, you know the human skin, right, that’s a hierarchical structure, from what we see at the macro scale all the way down to the cellular level, in multiple orders of magnitude, right. So at the microscopic level, wood is highly vascularized, contains long oriented strands, if you look at the the radial sort of section cut, as well as tangential, so you’ve got these sort of oriented fibers that maintain this really great flexibility, will also can resist torsion and allow for the the bending.

6:41 So again, my, what I do is I’m just like, look at something that like looks really cool, and I’m like, can we, can we model this, and what would be the point in that, and modeling the system in a way that’s available for somebody to use, like in a in a bracket, or in an implant, or something like that. So there’s there’s two different things, and we’ll look at the kind of the solution on the in the middle, right, the the left, and then the solution on the right, and what the difference between the two.

7:08 So I’m going to do something that no one ever recommends, and I’m going to model live, I’ve never been bitten by it before, tempting fate. All right, so here’s the thing, we got a set of boundary conditions, we’re interested in understanding how the how the load sort of interacts with the structure, and what the structure does. So here we just, you know, it’s simple push force, we’ve got one structure here, the thing on the left is actually a more common algorithm, and it’s basically says optimize my structure essentially keeping all the load conditions under whatever my thresholds are. But what’s interesting is you’ll notice something really cool, is on the right, and this is an algorithm that I think people need to do a little bit more exploring, and essentially says, in response to this particular loading condition, what’s the geometry that allows me to maximize displacement in that way, right.

8:00 So we’re, you know, when you talk about geometrical compliance, compliance from an engineering standpoint is, let me sort of hold the load with as limited geometry as possible, but I’m really more interested in the reverse, what allows me to maintain hyperflexibility in a certain geometry, because I do a lot with metal, right, how do you make metal flexible, what sort of structures do you need to create. So if we’ve got, you know, another problem, right, we’re going to hold it in the in the middle, and we’re going to smash both ends. I I like saying smash, I don’t know why smash instead of squash. And so the technicalities of the workflow are less important, because again this is not necessarily the time for that, but what’s actually really cool is the the, I I’m, I’m partially dumb, but I’m not that dumb, so I don’t, I didn’t include the iterative loop towards convergent, this is artificially looped only six times to get to a a false convergence, because I thought like if I was going to do that live, that would be really dumb.

9:00 So what’s interesting, look at the one on the, this middle one, versus the one on the right, this is a case where, like, we’re holding the middle, arm want to squish both ends, you’ll notice that the structures here are oriented specifically to resist, right, it’s columnar structures, whereas these have this sort of sandwich shape, and that maximizes displacement in that location, which again, from an engineering standpoint you say, well, you know, I I have got this load case and I want to be able to withstand whatever my maximum, but we query completely different things, we’re not really querying strains or stresses as much, because it’s going to self-repair, again, an entirely different way of looking at it, from an engineering problem, which is, just, I’m designing a single static structure that has to last for hundreds of years.

9:43 So then we can go to a separate problem, one that that, oh, again, we’ll share this real quick, so we’ve got this common algorithm, structural compliance, and then we’ve got biological algorithm, would you take structure and you just say compliance, I need it to move, I want it to move, it needs to be able to interact with its surroundings in a way that’s meaningful. And again, this is this is something that, again, I’m very familiar with in bones. So biological scaffolding, and most of the most of the time people, like, it’s really cool seeing what generative algorithms are doing, I I moved past the point of where I think, like, functional grading, like thickening or cell size is cool, to the point of, like, morphological grading, because I think that it provides a far more sort of complex interactive way of how we deal with the geometry, especially when looking at biological systems, which almost always are morphologically graded to their surroundings.

10:34 So quick background, if you don’t know, but basically in the 19th century this guy said, hey, bones are adapting to their surroundings in a way, Harold Frost came in the 1960s and defined it in this Frost mechanistat theorem, which is shown here, which is basically like this loading zone, which is the homeostasis, has a disuse where elements are trimmed or removed, and an overload where we get key growth, and then a pathological overload basically says, I’m being stimulated too much, and it doesn’t create the right bone formation. So you’ve got this perfect strain rate where you get this really cool sort of modeling scenario. What’s interesting when people, like, when you actually go and look at the structure of, like, a biologic scaffold, you actually get like weird stuff, disconnected elements, things don’t necessarily make sense all the time.

11:24 Again, from an engineering standpoint, you’re thinking, I need to maximize like the material that I have, I’m weight savings, and I need like this to withstand a load, and only use this much material. But biology kind of like, whatever, I don’t care about that, there’s just, it’s a different way of dealing with it, which again, then, from an engineering standpoint, it’s like, I need to solve this problem. So we don’t really think of that, again, the the disconnect and elements don’t matter biologically, because the, because again, there’s a, there’s a chemical reaction which basically says, this, you’re not getting used anymore, and it starts basically resorbing the bone back in the body, and using the minerals for something else, to get back to homeostasis. So you’ve got a maintenance, reinforced, and then a remove, that’s part of this biological homeostasis.

12:10 So I actually, this is when I started thinking about, you know, what if, what if an algorithm sort of did the same thing, and nTop actually gave this whole, like, this field optimization, and other people are starting to get this idea of, like, thresholding out the things that don’t need to be there, right, it’s topology optimization plus using it with an algorithm, but I will say, I’m going to show another thing, I was doing field optimization before field optimization was field optimization.

12:35 So the idea behind this is sort of an evolutionary algorithm, and it’s this idea that we, we don’t ever just look at a single time dependent step, but we look at a transient sort of flow from loading, and understand, like, if this did get implanted in the body, what exactly is this going to see over time. So specifically for, like, reabsorb-able materials, less so, is it, like, you know, this is a titanium scaffold, but really what we’re interested in and something that’s going to be in, like, ingested by the body over time, and how is the the part potentially going to change its shape. So we’ve got this sort of time dependent evolution, you can do stresses, strains, displacements, however you want to input the the sort of the problem, saying what’s going to happen with this thing, obviously simplified from a geometry standpoint.

13:19 And what you’ll notice is, like, if the algorithm is biological versus what you would consider structural, you get elements that don’t necessarily make sense engineering sense, right, they’re disconnected, they still follow some sort of structural compliance problem, but they don’t necessarily act in a manner which humans would say, I need just these trusses to connect the shape from top to bottom to get where I’m going. And I think that’s again what I’ve been particularly fascinated by, because it’s less so a problem of, again, current algorithms sort of want to converge on a solution, but when you look at, like, biologically, there’s not really any convergence. Because by the time it gets to the point where it’s stable, there’s already other influences acting on it, and it just needs to change again.

14:05 So when we look at load response, now we’re looking at something that’s more like strain response, so again, if you look at it from an algorithmic standpoint, and you look at each time dependent step, sometimes it makes sense, and sometimes it doesn’t. So when you just talk again about what something should be doing versus what, what it actually is doing, some, sometimes things make sense. So so basically, in this case, we’re just, we’re like tipping it over from the top, we’re going to push it, and you can look at the algorithm itself, and and you can actually sort of be like, all right, I can understand it, why this is forming to a certain extent, again, we’re prioritizing displacement, we’re not prioritizing sort of structural compliance in the way that most engineers define it.

14:46 But if you look at a, you know, a more complicated thing where we kind of have multiple constraints, again, because I’m doing it live, I’m not dumb, so I I’m only, I’m only going so far, and so we sort of have a little bit more of a load case where, you know, we’ve got the sharp geometry at the corners, and we’re kind of pushing and pulling in a certain way, what the solution that manifests, where we start thresholding out elements and sort of trimming things to shape, you start getting missing pieces. And again, from a, from a pure engineering standpoint, again, I’m an engineer, and I know all the like right rules for engineering, but what I’ve always been fascinated by, again, is these sort of systems, biological printing, printing materials that resorb, and that grow, and that change over time, I think are only going to continue to be more important.

15:39 The the end keynote yesterday was another sort of fascinating example of utilizing local materials, local manufacturing processes, to create something that’s just, you know, very ingrained with sort of nature, right, this this ecology, the sustainability factor. And similarly, again, with the body, my entire, the last 15 years have been spent talking about creating biologically relevant input plants. Biologically relevant means the best implant is one that the body doesn’t know exists, but that forgets it exists. So when we start talking about resorbable materials, and printing with bone itself into a scaffold that’s resorbable, you don’t, you can’t make the traditional assumptions that you used to make, which is, I got to take this amount of stiffness for x amount of time, and it becomes a more multifaceted problem.

16:20 So again, when we step through the slowly, there’s my timer, we’re out of time, when we step through this a little bit slowly, you can see sort of the evolution of the algorithm, again, exists solely in these sort of time dependent states, as, and not interconnected. In other words, sometimes elements go away, and it was the next step, and it doesn’t make as much sense to you from a human standpoint, but it’s just the evolution of what the algorithm sees and how it responds to the conditions that it’s been given over time.

16:52 So I mean, that’s basically it, you know, I really appreciated the opportunity to be able to to chat. Thanks to Juan so much, and hope to see you guys around.

More from CDFAM NYC 2023

Design Computation Human

Design Computation Human

Onur Yüce Gün · New Balance

Mud y Robots: Keynote Presentation

Mud y Robots: Keynote Presentation

Ronald Rael · UC Berkeley / Emerging Objects

Roadmap to Replicators

Roadmap to Replicators

Neil Gershenfeld · MIT Center for Bits and Atoms

Scaling & Standardizing Patient Specific Design

Scaling & Standardizing Patient Specific Design

Adam Wentworth · Mayo Clinic

Computational Design Ops

Computational Design Ops

Ronnie Parsons · Mode Lab

Overcoming the Design Dilemma

Overcoming the Design Dilemma

Ryan Carter · Johns Hopkins APL

Generative Design & Digital Manufacturing at NASA Goddard

Generative Design & Digital Manufacturing at NASA Goddard

Ryan McClelland · NASA Goddard Space Flight Center

Digital Materials – Geometry and Data Combined

Digital Materials – Geometry and Data Combined

Marek Moffett · General Lattice

Ten Years Teaching Additive

Ten Years Teaching Additive

John Hart · MIT

10 Years of Applying DFAM

10 Years of Applying DFAM

Tim Simpson · Penn State University

MfAM Walks so DfAM Can Run!

MfAM Walks so DfAM Can Run!

Chelsea Cummings · The Barnes Global Advisors

Informatics and AI for Materials Development

Informatics and AI for Materials Development

Wes John Alder · Citrine Informatics

Predicting Fatigue Life in AM Metal Parts Using DigitalClone

Predicting Fatigue Life in AM Metal Parts Using DigitalClone

Behrooz Jalalahmadi · Sentient Science

Designing Material Properties with Advanced Toolpath Generation

Designing Material Properties with Advanced Toolpath Generation

Alex Roschli · Oak Ridge National Laboratory

Design Automation Using Machine Learning & Computational Design

Design Automation Using Machine Learning & Computational Design

Rhushik Matroja · Cognitive Design Systems

Learning to Generate Shapes

Learning to Generate Shapes

Karl Willis · Autodesk Research

Design for Artificial Intelligence

Design for Artificial Intelligence

Chris McComb · Carnegie Mellon University

Register for Updates and Discounts on CDFAM events.