sign in sign up
<<
Home , Generative design, Product Design

XXII Generative Conference - GA2019

Applying Generative Systems to Product

Prof. Alex Lobos Department of , Rochester Institute of Technology, USA www.rit.edu/artdesign/directory/aflfaa-alex-lobos e-mail: [email protected] ______

Abstract Author’s design explorations where both approaches for are used in . Examples include Generative Design provides multiple products using Voronoi patterns and benefits to the development of new procedural networks where the physical products. First is the creation of intricate appearance of the product is strikingly patterns that resemble natural systems, intricate and appealing, while the physical moving away from geometric shapes attributes of the product are not typical of mechanical design. Second is necessarily improved. Other examples the automation of processes where illustrate the application of generative computers perform complex and repetitive design structures created freely by the tasks that would be too hard or tedious for computer, following only set goals for to do. The opportunities that supporting weight loads at given points. automation provides is frequently This process results in unique structures considered the main benefit of generative that are lightweight and strong but might design in the creation of new products, also have a polarizing appearance for buildings and systems. In both of these specific product applications. These approaches, the output that computers examples will enable discussion on how generate is driven primarily by a will continue to integrate ’s vision that already has a automation and generative systems into general idea of how the result might look their process as technology continues to like. A new approach for generative develop. design by software company Autodesk allows designers to define goals and criteria for functional CAD , and then having a program generate iterations 1. Introduction of potential solutions. This process presents a radical shift where the Generative systems have been long used computer is not just facilitating the ideas by various civilizations throughout history. of the designer but rather designing itself. Examples of geometric studies related to While designers still are in charge of the astronomy and the are found in process, deciding which solutions are ancient Greece. Islamic art is perhaps the suitable for further refinement and best examples of how humans use implementation, the relationship between geometry to understand the relationships and machine becomes between humans, nature and the divine collaborative. [1]. The use of computers in the 1960’s to generate complex geometries was the This paper explores the concepts origin for generative art [2]. A couple of described above and it shares the decades later its use expanded to in page 1

XXII Generative Art Conference - GA2019

Architecture, and This paper provides insights of the Construction (AEC) industry, leading to workflows describes earlier, and uses what is known today as generative design design explorations developed by the [3]. As digital computing has become Author, to illustrate the use of different prevalent in our society, automation is types of generative design strategies in now a key component for generating industrial design. These workflows are just complex solutions such as the ones in a few of many ways of integrating which generative design is based. generative design into creative processes Automation reduces the burden of for product development. As designers performing repetitive tasks, allowing become more familiar with these people to focus their energy and intellect workflows, generative design will into more meaningful activities that consolidate itself as a key component of improve quality of life. The benefit of design process, that helps in developing automation goes beyond just performing solutions that are efficient, visually repetitive tasks, opening the door to engaging, and better aligned with natural achieving complex tasks that would be systems. impossible for humans to perform by themselves.

Generative Design is a type of artificial 2. Generative systems from intelligence that develops unique shape tools to collaborators grammars based on three types of information: a set of rules for rules, a way Designers benefit greatly from generative for shapes to develop, and goals to be design and its ability to handle large and meet by the resulting shapes [4]. Goals complex amounts of calculations. While can include factors such as mass, designers need to understand the basic strength, or manufacturing processes. relations that systems need to have in Generative design provides several order to create a particular pattern, they benefits, including the development of do not need to have full knowledge of how shapes with unique geometries that are to perform these calculations in their aligned to their context and improvement entirety. By providing basic sets of rules of mechanical performance [5]. and objectives, computers are able to turn them into that create 3D shape Industrial design is embracing the grammars [7]. There exists a wide variety principles of generative design, first of software programs that perform these focusing on the unique aesthetic style that tasks, and two of the most popular today comes from nature-inspired organic for AEC and industrial design are shapes. Designers have also been able to Grasshopper and Dynamo. These integrate automation into their process, programs work on brackets of information which provides a streamlined workflow that generative complex calculations. and intricate geometries with the potential Designers need to have the knowledge of being efficient and resilient, providing necessary to input commands to the advantages in mechanical performance program, and also to program them in with and use of materials [6]. The integration of the right commands so that the results are generative design can happen either as a successful. Once this is accomplished, the tool that executes the vision that hard work occurs “under the hood” and designers have for a given product, or as the computer takes care of performing the the creation of novel forms that meet tasks that have been assigned to it. The goals set by design problems. key element in all of these benefits is that

page 2

XXII Generative Art Conference - GA2019

generative design is a powerful tool, that Cellular Automata, Swarm Intelligence, complements and enhances the and Generic Algorithms [9,10]. Each one capabilities that designers have to create of these categories contain variations complex systems. within as designers and artists experiment with algorithms and geometries in novel Generally speaking, generative design ways. In this paper, three types of can be used in product development in generative systems will be analyzed and two ways: one is as a tool that performs described: Voronoi patterns, procedural calculations in order to obtain a networks, and generative structures’ preconceived solution that the designer software. These three systems have been needs. In this case automation focuses as selected for the interesting way in which a tool that optimizes the creative process. shapes are created as well as because of The second way in which generative how they complement the traditional design can be used is by providing it with process that industrial designers use to goals and parameters that a given develop new products. The goal of this solution needs to have, and letting analysis is to encourage designers to computers generate solutions integrate generative systems into their autonomously, without any preconceived current workflow, or to expand their notion of how a solution might look like. current use with new approaches. For each of these methods described within, This process elevates the role of the examples of products designed by the computer from just a tool that performs Author will be used to illustrate how tasks to an active participant in the design process and CAD software can be creative process. The computer is now used. more than a tool for the designer; it is a collaborator that provides solutions that would had been unimaginable otherwise. Generative design becomes a source of 3.1. Voronoi patterns that is evocative and adaptive [8]. Its evocative character is based in its Voronoi patters are some of the most ability to evoke new thinking and to common generative systems used in display new ways of solving problems. Its product design. They are subdivisions adaptive character is shown in its ability to applied to a plane or surface that sit at the generate solutions in a wide variety of same distance from a specific point [11]. applications. Voronoi patters are quite popular in industrial design as they can be created with just a few, straight-forward parameters: area to be covered by the 3. Generative methods in pattern, distance between cells, and point product design of origin, called seed. There are several online applications that automatically The are several types of generative generate Voronoi patterns and several systems, all of which provide varied CAD programs also include this tool either workflows and resulting shapes. as a native feature or as an add-on Generative art is an excellent starting extension. Voronoi patterns are visually point to analyze and understand how engaging, with the cells distributed in a generative systems and grammars work, fluid order. They are commonly compared and the shapes that they produce. There to the interior mesh of bones, and in are five main categories of generative general they highlight organic, intricate systems: Shape Grammars, L-Systems, nature that generative design is page 3

XXII Generative Art Conference - GA2019

associated with. A key characteristic of (See Figure 2). The bass guitar was Voronoi patterns is that they are based on originally designed as a solid shape. After pre-existing designs. There is no need to the shape was completed, the main body learn complex algorithms in order to of the instrument was divided into generate shapes. This setup provides sections. The outer sections were turned designers with ample control of the design into a Voronoi mesh using an online process, which takes a solid body and shape creator called Voronator.com. The cores it out with the Voronoi pattern, sections closer to the electronic similar to how a filter is applied to a 2D components remained solid, in order to image (See Figure 1). not affect the functionality of the base as well as to provide enough solid material for the bass to sound well and avoid any . A similar exercise of keeping solid sections was applied to areas that connected with the neck of the bass guitar. The result is an instrument with a strong visual aesthetic and that is significantly lighter than its solid version.

Figure 1. Comparison of a solid body and its Voronoi-based version.

Along with their visual appeal, Voronoi patterns can generate forms that are stronger and lighter than their solid versions. By turning solid surfaces into meshes, the geometry of the bodies is filled with more edges, which provides more strength and stability. The mesh at the same time removes surface area, which depending on the manufacturing process, can reduce the amount of Figure 2. Electric bass guitar with Voronoi material needed substantially. When 3D body. printing the bodies shown in Figure 1, the The same workflow was used to create Voronoi version would use 40% less the stool in Figure 3, which combines solid material than the solid version. A potential sections with Voronoi-meshes. This trade-off is that many shapes that would allowed for more control of the shape of be fabricated with traditional the piece, making sure that there was a manufacturing processes would not be good flow and proportions all around the able to be created if they have a Voronoi object. The inferior sections of the stool version, due to the more complex were meshed out with a Voronoi pattern in geometry. For many Voronoi-based order to create a more interesting design. shapes, particularly if they are non-linear, The top section of the stool remained the only fabrication method possible will solid, in order to make the piece more be additive manufacturing, which could comfortable to sit on, avoiding increase fabrication time and cost. unnecessary pressure points that could touch against the body. In this case, using An example of a Voronoi pattern in the Voronoi mesh in the inferior parts of product design is an electric bass guitar the stool adds mystery to the overall page 4

XXII Generative Art Conference - GA2019

design, creating curiosity in the user, geometries, ranging from random patterns without compromising the performance to more controlled progressions. Patterns and the perception of strength and created in MASH can be exported in FBX stability than a seating object must deliver. format and imported in most CAD programs. The use of this workflow is innovative by itself as CAD programs like Fusion 360, Solidworks, Inventor, Rhinoceros, etc., which are common for product design applications, do not provide simple ways of creating complex shape patterns such as procedural networks. A limitation of the process to keep in mind is that Maya generates polygon-based geometry. This type of geometry is not parametric and depending on how complex it is, might result in a very dense model to import into CAD programs because of all of its vertices and edges. Figure 3. Stool that combines solid and Voronoi-based sections. Figure 4 shows an application of procedural networks that were developed 3.2. Procedural networks in MASH and then imported into Fusion 360. A grid of rectangles influenced by a Procedural networks are widely popular in noise signal, which creates a random Generative Art. They are based on code effect with a strong visual appeal. This segments or algorithms that define effect is applied to a block of translucent physical characteristics of a CAD model material, such as glass or acrylic. When [12]. The benefit that they provide is that applying light to the translucent material, instead of fully defining forms, textures or this creates a very interesting lamp, that animations, which would require massive takes advantage of how the light refracts amounts of data, they define the coding of at different levels along the edges of the how these physical attributes will be structure. generated. Procedural networks are very popular in Media and Entertainment for their ability to define the way in which body particles will move in and be affected by external forces such as gravity, wind or other objects.

The use of procedural networks in industrial design, however, is fairly limited. A reason might be that industrial designers tend to design static bodies that only move when being used by/for the user. In order to explore the benefits of Figure 4. Table lamp with shade using a procedural networks in product design, a grid of boxes with a random pattern. series of models were developed in

MASH, a procedural plugin part of The use of procedural networks is applied Autodesk Maya. MASH quickly provides in a unique way that benefits industrial exciting and powerful results for complex page 5

XXII Generative Art Conference - GA2019

design. Procedural software such as manufacturing, making the process of Maya MASH, was developed for design, simulation and fabrication, a lot animations. This implies time-based more integrated and streamlined. movement that makes the patterns move and transition between positions. For their In 2018, Autodesk released their use in products, designers take a Generative Design application, which is “snapshot” of one of the frames in the part of Fusion 360. This tool works sequence and use that to define a static radically different to programs like shape for an object. This is one of many Grasshopper or Dynamo, and it brings the applications, of course. There can be very process even closer to what true exciting cases in which products have generative design is. Based on artificial articulated parts and they can take further intelligence, this program takes the set advantage of the procedural sequences, rules and goals from the user and creates envisioning a wide range of movement. large amounts of shape grammars from There are even insightful scratch [15]. This process is different from experimentations on using motion time- generative patterns such as Voronoi, as it based procedural modelling to create doesn’t need a pre-existing shape to base complex bodies from their steps as they its form, and it’s also different from move through space and time [13]. procedural modelling, as it takes data from the user not as directions for creating shapes, but rather as a wish list of goals to be met. With the Generative Design 3.3. Generative structures app, the user provides data for areas or shapes that need to be preserved as well Generative Design has captured the as well as obstacles that should be imagination of product designers for avoided when creating a new form. More decades. Up until recently, most importantly, goals are set so that the designers who integrated nature-based design is able to support specific loads or patterns into their workflow did it in a way physical forces. Once the inputs are that was fairly manual. This means that complete, the computer takes this they would generate individual shapes information and automatically generates multiply them in patterns with incremental multiple designs, all of which solve the variations of shape, position or scale, in design problems in a unique way. The order to achieve an organic transitional longer than the program runs, the more flow. Programming software allows for iterations that will be produced by the users to develop sets of rules and computer. The user then goes and checks behaviors that turn into interesting shape on the results, which vary in form, material grammars. The high knowledge needed used, strength, and manufacturing for successful programming has limited its processes, although all of them meet the adoption across design fields but basic criteria set by the user. Designers programs such Grasshopper and Dynamo are able to pick from these solutions the have found effective ways for making ones that they feel are better for the programming more accessible. What application, and they can integrate them makes programs like these better, is that into their design, or use them as a base the programming is visual, based on for a new iteration of generative design brackets, connectors and widgets, making simulations. This process is novel in the generative system easier to product design for several reasons: understand [14]. Additionally, these programs are easily connected to other  The creation of forms is CAD applications used for design and automated, which is one of the page 6

XXII Generative Art Conference - GA2019

key goals of generative design. This process allows for large amount of data to be processed by computers, removing the burden from the user. In many cases, processes like this are limited when run by humans, because of limited time, knowledge, or interest in performing large amounts of calculations, repetitive tasks.

 The amount of solutions that are

generated by the software are Figure 5. Ceiling lamp combining virtually unlimited. This provides generative design and procedural a rich foundation to designers to networks. pick the perfect solution, instead of being able to generate only a Figure 5 shows a ceiling lamp created in few iterations, not knowing if they Fusion 360. What makes this lamp unique have found the right one. is that the actual modelling part of it is minimal, and in fact it didn’t have a  The process provides an significant impact in the final shape. Most increased of creativity to of the design was created by using the designer. The computer is generative tools within and outside of able to create solutions that many Fusion 360. The base of the lamp was times would had been created using the generative . inconceivable to designers, simply In this case, five points were assigned in because human brains don’t space: one up high to define where the process information the same way lamp would be hanged from, and four that computers do [16]. point that would connect the lamp to the Additionally, designers can be shield. At each of this point, forces were inspired by the results that they defined in terms of weight load and obtain by the software, pushing moving forces to make sure that the their creative process to even structure would be strong enough to hold higher and broader levels. the weight of the lamp as well as external forces such as someone hitting the lamp  With generative design software, accidentally from the side or below. A few the computer is no longer just a areas were also defined as “obstacles” to passive tool that follows orders of make sure that the resulting shape remain the designer, but it becomes a within a certain envelop and didn’t expand dynamic problem solver that too much. The results of the simulation provides unique design solutions. provided several exciting possibilities for This is perhaps the more the shape of the lamp. One of the substantial contribution of solutions stood out because of having a Generative Design to the design very unique aesthetic that seemed to have field. The computer becomes not a good balance and good potential for only a tool for the designer but an looking like a “lamp base”. The shade of actual collaborator. the lamp was created in Maya MASH. A simple cube was created in the workspace, and a series of procedural page 7

XXII Generative Art Conference - GA2019

nodes were applied to turn the cube into a is becoming a key element of industry and three-dimensional grid with a progressive designers need to become familiar with it, deformation. These effects resulted in a and to make sure that it’s used in a progressive pattern that is both geometric meaningful way that enhances human and organic, and that creates an productivity, rather than just replacing it. interesting contrast with the lamp’s base. As generative design matures and Once the shade was completed, it was becomes a common component of design imported into Fusion 360 and combined process, it will be easier to find the right with the base and lighting fixtures as a balance of and performance, traditional CAD assembly. so that products out in the market are embraced by consumers while also providing the efficiency that natural systems have. 4. Conclusions

Generative design enables an exciting direction where different workflows and References ways of achieving nature-inspired systems can be combined in new and unexpected 1. Dabbour, L. (2012). Geometric ways. In the case of the lamp shown in proportions: The underlying Figure 5, the design combines two ways structure of design process for of creating generative systems. The Islamic geometric patterns. shade uses procedural networks to create Frontiers of Architectural a dynamic pattern with an interesting flow Research 1(4): 380-391. an elegance. This approach shows a 2. Boden, M., Edmonds, E. (2009). process where the designer had a general What is Generative Art? Digital idea of the effect that the shade would Creativity, 20 (1-2): 21-46. have and used digital tools to achieve this 3. Frazer, J., Frazer, J., Liu, X., effect with ease. The base of the lamp Tang, M., Janssen, P. (2002). shows a completely different way of Generative and Evolutionary generating shapes, which was based just Techniques for Building on goals for supporting weights and forces Envelope Design. In: Soddu, C. at specific points. As generative design (ed) Generative Art 2002: 5th continues to evolve, it integrates criteria International Generative Art for shape development and optimization Conference. 11-13 December, with goals for engineering performance Milan, Italy. [17]. The result are shapes that resemble 4. Krish, S. (2011). A practical natural structures as much in the way that generative design method. they look as they perform in living Computer-Aided Design 43: 88- environments. 100. 5. Agkathidis, A. (2016). Generative Generative design is becoming a key tool Design: Form-finding techniques for industrial design, providing a wide in Architecture. Laurence King number of benefits. The most important Publishing, London, 10-14. one is the increased capacity that 6. Lobos, A. (2018). Finding automation brings to the design process, Balance in Generative Product whether it is as a way of avoiding Design. In Proceedings of repetitive tasks that become monotonous Norddesign Conference: Design and time consuming, or limitations to in the era of digitalization. 14-17 perform complex calculations. Automation August. Linköping, Sweden. page 8

XXII Generative Art Conference - GA2019

7. Tang, M.X., Cui, J. (2014). Generative Art Conference. 12- Supporting product innovation 15 December, Ravenna, Italy. using 3D shape grammars in a 14. Ferreira, G. (2016). An enhanced generative design framework. programming environment for International Journal of Design generative design. Master thesis. Engineering 5(3): 193-210. Instituto Superior Técnico. 8. Avital, M., Te’eni, D. (2009). Universidade de Lisboa. From generative fit to generative 15. Autodesk (2018). What is capacity: exploring an emerging Generative Design. Accessed dimension of information July 25, 2019. and task https://www.autodesk.com/soluti performance. Information ons/generative-design Systems 19: 345-367. 16. Khan, S., Awan, M.J. (2018). A 9. Singh, V., Gu, N. (2012). generative design technique for Towards an integrated exploring shape variations. generative design framework. Advanced Engineering 33 (2): 185-207. Informatics 38: 712-724. 10. Zimmerman, L., Chen, T., Shea, Oh, S., Jung, Y., Kim, S., Lee, I., Kang, N. K. (2017). Generative Shape (2019). Deep Generative Design: Design Using 3D Spatial Integration of topology optimization and Grammars, Simulation and generative models. Mechanical . In: Gero, J. (ed) (upcoming edition), 1-22. doi:10.1115/1.4 7th International Conference on and Cognition. 27-29 June. Evanston, Illinois, USA. 11. Lobos A. (2018). Next-Level Design Workflows in Fusion 360: mixing T-splines, meshes & Voronoi-based patterns. In: Proceedings of Autodesk University. 13-15 November. Las Vegas, NV, USA. Accessed July 19, 2019. https://www.autodesk.com/autod esk-university/class/Next-level- design-workflows-Fusion-360- mixing-T-Splines-meshes-and- Voronoi-based-patterns 12. Ebert, D., Musgave, F., Peachey, D., Perlin, K., Worley, S. (2003). Texturing and Modeling: A procedural approach. Morgan Kaufman: San Francisco, CA, USA. 1-4. 13. Obradović M., Marković, S. (2017). Creating 3D shapes by time extrusion of moving objects. In Soddu, C. (ed) Generative Art 2002: 5th International page 9