Generative Design in Nike’s Innovation Department | Lysandre Follet

Hope you had a good break maybe a short walk. It’s a beautiful space So I’m ready own out to be here in San Francisco today and my name is Lysandre Follet And I’m the director of generative design at Nike where we sit within the innovation design studio so 1968 Doug Engelbart on the group of 17 researcher working with him in the augmentation Research Center at Stanford University presented a 90-minute live public demonstration Called analyst the analyst system was really the first to employ the practical use of Mouse Textual information and it was really a groundbreaking event. It’s been cold since then The mother of all demos and people keep referring to it What is interesting on that I want to push on is at the time in Stanford. It was also a lab Imaging lab working on artificial intelligence. But what Doug group was really focusing on was augmenting intelligence In 1962 Doug Engelbart said that a new tool doesn’t just make something easier. It allows for new previously impossible way of thinking of living of being 1962 Nineteen nineteen seven after a few years of research on development IBM built deep blue a state-of-the-art computer Designed to run powerful algorithm to play chess against the best player in the world I think we all remember what happened in May 11 97 This big computer deep blue is set for an historic chess game with legendary chess player Garry, Kasparov on that day Something happened d blue beat the world chess champion after a six-game match two wins for IBM one for the champion But what a lot of people s people know about is that following is defeat Against the blue garry kasparov recognized the potential of such an advanced computer system Not just as a human versus computer setup But as a partner in the game that idea of partnering with the computer Jerry just power of developer version of chess called advanced chess also sometimes Referred as Center chess, which is an interesting term The objective was ready to have a human player on a computer chess program playing as a team That notion of team to or not trying to beat against the computer but we were playing together to and ends the game on many advanced chess player stresses that Playing as a team With a computer as married in increasing the level of play – I’d never before seen in chess also producing Brander free game with the qualities on the beauties of both perfect tactical play on Ally meaningful strategic plans on giving the vewy the viewing audience an insight into the thought processes of strong human chess player on strong chess computer and not Like early adopter that idea was back then in 97 We all recognize today the need for light more transparency in learning experience, especially with artificial intelligence On the same year a couple of bright minds from Stanford lunch, Google We all know what it is now But at the time it was that simple idea of indexing web pages with the ranking algorithm At the time price was talking about complicated Mathematical analysis on John mountain from The Observer in 99 right about an ingenious Algorithm and that notion of algorithm is important on as evolved a lot over the years and I always loved that slide That show it’s a bit outdated but a really show from the simple search Algorithm the product has evolved with more and more feature on in the recent year a lot of intelligent feature So that idea that that piece of algorithm that ran on that computer can get more and more powerful Into the world algorithm artificial intelligence are everywhere in your pocket driving your car Unlike never before digital and physical world are already becoming linked on its transforming our we make on interact with thing What I’m interesting to talk about today is how we can leverage the same very concept that with just so powerful concept on Method to radically change the way we create product also are those product are becoming more and more personalized? So at Nike base in Oregon founded in Oregon from the beginning we’ve been maker craftsmen tinker on innovator it’s a mindset that really helped us push the boundary of what’s possible on change the landscape of modern spot design and This is Bill Bowerman. And in the late 1950s Bill Bowerman was really dissatisfied With available running spike. They were made of waiting later on metal as a result It really became obsessed with shaving ounce of shoot Well pronounced slash second of the best time. So this quest really one of Redefining athletic footwear and in the 1980s. We see here Bill Bowerman Custom making custom shoe for is athlete by right by often using like really analog meter They will draw the outline of a feet it will measure the width It will not individually these from all these different athlete so channels as an extended ill or slim ankle so that idea of Personalization tuning an existing product to make it better and more person personalized Nike mission statements and vision is to bring inspiration on innovation to every athlete in the world and Not only for professional athlete but all the athlete on as coach. Barman will say if you have a body you are an athlete So question we could ask is how do we design for every athlete in the world? well it’s really complicated because we all have a different body and we already are dealing with a complex design problem right there and Almost have all of you actually You can check to night before going to bed You will most likely figure out that you’re left on right foot are even different So I do will do is that I really like that visualization because it kind of give a good idea of the Burien diversity of body for a very different spot on the muscle group on the body that also to those Different activity and for the last few decades as we design product we have optimized against sample size so what we call sample size is often size men’s 10 or Women’s 8 and then we scale on modify as we go to on the way performance really isn’t perfected With tradition and method it kind of is a good as good as you can get So into a 13 we started working with Jamaica and Shelley and Fraser price printer She’s already I cadence Paris printer one of the best females printer in the world and the goal was to build for out the best performance bike for the 2016 Rio Olympic game on it She really was a good test subject for our program because she was wearing a size five spike, which is really really small feet And we knew that we had a lot of performance issue with that product. So We spent a considerable Amount of time to meet with athlete we listened to them and really put them at the center of our design process on using state of the art Equipment we have taken the data from thousand of athlete in the last decade from the best marathon athlete to someone that just started running And here is our despot science collect the data back on companies in Oregon if you take science and design by themselves They have no power They’re not going to do beautiful things, but the overlap and the convergence of those two together as well beautiful things happen It’s an extremely creative process to be researcher in the United score search lab If you look at all of us We’re all extremely passionate about sport we geek out on the science of athletes and athletic performance Every day is a new challenge and we want to make athletes better all the time Athletes asked us for three things They asked us to make me better to protect me and to inform me The better you understand the needs of that athlete the better you’re going to make product Motion capture is one of the tools we use what it allows us to do is really objectively understand an athlete in motion We use high-speed video most digital cameras capture around 30 frames a second We can do 30,000 frames a second so we can see things you don’t necessarily see with the naked eye our force plates take thousands of measurements per second and we get forces in three-dimensional space we Have environmental chambers we can collect things like expired gases So we know the amount of energy your body is using to do a particular movement and in these chambers We can accurately replicate weather from from anywhere on the planet and make sure that our product is going to work in those environments So we have a lot of data but with traditional method it’s often hard to make sense of it So how do we best leverage data to inform design? well most of the creative software for the last decades have been replicating traditional processing giving the experience of drafting drawing or sculpting It translate our human have traditionally been working with physical material So at Nike design has always been highly iterative with Aidid dreamed up on Evaluated refine or discarded on the way to the kind of innovation that define an industry So as the athlete we self continue to break barriers We have to keep pace on we have to stay ahead get smarter faster and better so in 212 we started to look at a Emerging design method on process called generative design and I always like to go back to the definition from Wikipedia That say that generative design is the application of computational strategy to the design process On that the goal isn’t to document the final results necessarily but rather the step required to create that results so well with Traditional design you will kind of like maybe sketch something and move into the process and have one or couple of solution with generative design What is interesting is you really want to document the step to create all the results that will be possible? So generative design really changed or the practice of design is expressed from simple drawn geometry to a least sophisticated Algorithm generated logic. It really is a revolution of problem-solving generative design has been every use in the last decade by architect with by owners such as I Did and it’s not transforming a lot of industry from medical to car company On those are often the three element that you will find in a generative design process Which is design schema a means of creating variation on a mean of selecting desirable outcome So let’s unpack that a little bit so step one We saw that data collection is very important on it it is even more opponent in a generative design process and we’ll see later why so in 213 we but Shelleyan in the sports research lab in Portland on by prints placing sensor in the shoe We were able to track how much forces cheryan apply on each stride So once the researcher gathered that data and finish to analyze the data we started to have a good understanding of what needed to happen To make a good it performance target and what we already wanted to try to do is to be able to create a tunable Spring plate structure. There will be as light as possible while maintaining an accurate control on localized stiffness So the step two, which is maybe one of the most important step in that Generative design method is Tran selling design problems into computable parameters Or we could rephrase it as saying Tran selling the design problem into constraint in the generative Objective what we did is we look at nature for guidance and we found this photo synthesizer It’s a prank tonic Chatham’s It’s a microscopic organism that live in the ocean and that service problem really well is very light but it has a stiff shell so this organism has evolved over millions of years to become one of the most performing a system in his group of ptah plankton So the idea here is not to just take the aesthetic and bring it back to a product It’s ready to look for the principle on NASA tech chure of that organism on This is how we study to shape the design process by writing algorithms that will define how the design get constructed It is that interesting idea of building a tool versus using a tool so we’re not using an out of the chef’s tool We’re actually building or own tool like some Maybe a lot of people were doing before in woodworking Well, they will need a specific tool on they will build it for a specific purpose. So we’re writing algorithm and of course, it’s like a digital tool, but it’s a tool that Again doesn’t just make something easier but a love for a new previously impossible way of thinking creating of making for us So the step three is creating variation once we have a system in place We can quickly generate massive amount of design We can experiment with parameter make adjustment to the algorithm on introduced constraint on the designer on the algorithm really start to work together To translate this understanding to create something extraordinary a design never before seen with a mathematical simplicity What next this freed geometry unlock a fascinating kind of living intelligence One that treats those Paranal structure not in a static spreadsheet, but as an active eco system It’s a constant moving system It is then translated into a variable stiffness a structure similar to the one fine on the geums This structure can provide more support where more support is needed for each individual. Yes I’ll eat you enable grant contact traction on an optimized wait – stiffness ratio the step four is a very Exciting one for a creative person which is exploring design space. So we talked about that idea that we don’t create just one solution, but Non finite number of solution and that’s that. I did a generative design offer a very different approach We’re leveraging dynamic system that we can interact with So instead of creating again that static design we build a dynamic system each time We run a loop of that algorithm. We generate thousand of solution building a design solution space a non finite space you know, we see a design exploration fraud Ronchetti by software company Autodesk based in San Francisco actually will Recommend everybody to go check out their facility In San Francisco, it’s pretty cool to have a nice gallery on what they are doing here is a multiple design solution They are exploring for that drone chassis. It gets optimized by the computer to find the best solution possible to the problem It’s endless possibility. We really see that we can unlock imagination Step 5. Of course, we now have that huge design space. We need to find a way to search on curate that design space and We have a couple of different method One of the thing that we do a lot is we leverage computer simulation to validate the design solution performance target it’s really fast on yellow dust to quickly understand what design space we can narrow down the design space the search space and Remove all the solution that will not meet the criteria So here we see bending stiffness test or the solution that will not pass that test Get removed and we start to understand which subset of the design space we need to focus on The second thing we do is a lot of real-world testing Leveraging 3d printing and additive method we can drastically cut development time on expensive tooling on produce highly tuneable performance Footwear to test This is shelleyan first 3d printed prototype drawn on a Samba in Portland, Oregon in a matter of day We can generate and build a new design which before would have take us maybe one two Or potentially three months who sent to Asia have a mold made there on a path back And by the time we will get the path, we had already moved to the next iteration So now we are able to print them on assemble them overnight We can test them on athlete’s feet get our new data point on Feed that new data back in the system on rerun a design loop. It is Nelly iterative design process the Role of the designer at this stage is to reframe subjective human preferences such as comfort performance criteria combined with aesthetic Augmented by the computer the designer really take on a role of curation I like this one, but I think that they perform better This one is really ugly. And maybe this one is the it’s the one we want to go with With generative design, we have solved one of our biggest Problem, which is that idea of designing for the size So here is Freud lead they all run the same distance and red mineral spring They all world-class athlete the differences is they all have different shoe size and remember I told you that we know we in history We designed for sustained on then scale up and down what we did this time is going back to the algorithm understand What are the performance target for each size on rerun a unique? expression a mutation of that algorithm of that DNA and this gave us Multiple unique design as you can see the cell arrange and find the best configuration possible to meet the performance criteria So this was really a groundbreaking Groundbreaking idea and our realization for us and we didn’t stop there So again for world-class athlete and they are all running. The differences is they’re running from a hundred meter So straight line and then they start to take the curve around the stadium Up to a 10k with Mo Farah We have here four different distance. So why not doing the same exercise? So we based on the data We knew what needed to be done in terms of performance for each product And again, we mutated that algorithm on rerun specific solution So as you go from running a straight line to taking the curve the geometry get tuned Specifically to allowed medial to lateral flexion and for the 10k it gets as minimal as it can be based on the data providing where the traction and stiffness need to be On this happen after three years of art walk We could have not dreamt of a better stage Than Rio Olympic to validate on the track the performance of a product on the feet of most of the best athlete in the world Helping them achieved a dream. We had more than 45 medals in this new product on what we decided to do after that success is looking across the business and we say well we also have other category of product that are trying to solve that same Kind of like performance criteria of having a light product with a very highly tuned control over stiffness So we were able to leverage that very same algorithm to create solution across a category from American football but also baseball on what we have already done here is Creating a family of performance product that all share that same DNA that same Algorithm that mutate to create those very specific Individualized solution, but it is from that same rules. It’s a familiar class of solution Relative process the computer generate thousand of design solution that the designer on its own will have never envisioned as well as many artifacts creative artifacts along the way Imagination is truly unlucky with surprise and delight the computer become a partner in creation It is co-creation and it is that very same idea that garry kasparov add in 97 with adventures playing as a team. We are creating as a team We are augmenting creativity and this is what is really exciting for a creative practice Machine on system are becoming more and more intelligent There is a beautiful learning opportunity for the designer on the creative process Today I wanted to finish on a quote from Lisa Dell after playing against on being defeated by alphago Alphago showed us that moves human may have thought are creative who are actually conventional I think this will bring a new project to go Thank you

One Comment

  1. JP jay said:

    subtitles pls!

    August 14, 2018

Leave a Reply

Your email address will not be published. Required fields are marked *