Analysis on Novel Feature-Independent CAD Design/Modelling Approach Combined with a Neural Network for Random Parametric 3D Boundary Representation Modelling

Analysis on Novel Feature-Independent CAD Design/Modelling Approach Combined with a Neural Network for Random Parametric 3D Boundary Representation Modelling | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Analysis on Novel Feature-Independent CAD Design/Modelling Approach Combined with a Neural Network for Random Parametric 3D Boundary Representation Modelling Nikhil Chaitanya Angajala, Srinivasa Prasad Balla, Rajeswara Reddy Resapu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3126805/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract In the era of competitive digital innovation, product manufacturing companies need rapid customisation and ability to create uniqueness in new product development to stay competitive in the consumer market. Till recent times, this requirement heavily depended on CAD designer’s ability and experience to produce creative designs. This paper presents a novel feature-independent CAD design/modelling approach combined with a neural network that enables the creation of random parametric 3D CAD design variants using the Boundary Representation (B-Rep) method. This method is rapid and thus offers about 10 to 100 concept alternatives to the client in 10 to 30 minutes. Additionally, the paper also highlights the suitability of the proposed Neural Network method in creating 3D deep learning datasets to train generative design models like 3D GANs to further enhance 3D designs specifically targeted in product design problems. Parametric Modelling Fusion 360 Deep Learning Generative Design Product Design Neural Network Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction A recent (2018) Harvard Business Review article [ 1 ] stated that although CAD (Computer Aided Design) was introduced in late 20th century to aid designers and engineers in creation of cad models, they significantly lack the “computer aiding” element and merely support only the documenting framework of what the designer/engineer is imagining in his head and the article suggested and predicted that Generative Design (GD) methods will be the next big wave in transforming the design approach across various engineering domains. As far as the GD modules available in present-day CAD softwares like Fusion 360, the design alternatives are based on the specified constraints like (i) Geometric Constraints such as obstacle geometries and geometric regions to retain, (ii) Structural Constraints specified in terms of load, factor of safety and mass, (iii) Material Constraints like Tensile Strength and (iv) Manufacturing Constraints like process restrictions that are specific to the manufacturing process like Machining, Die Casting and Additive Manufacturing. Although GD is a big step towards utilising Artificial Intelligence (AI) to create design solutions, it lacks the ability to generate aesthetic models and also lacks creativity in terms of possible design solutions that are specific to consumer product design such as design of outer body of electronic gadgets, body panels of automobiles, surfaces of cosmetic packaging etc. Hence these gaps form as the basis and motivation for the present study. Parametric Modelling refers to the methodology of defining the design variables (or otherwise known as parameters and hence the name ‘Parametric’ modelling) that store the values of dimensions and constraints besides defining other mathematical relations that govern the behaviour of the model that make up the CAD model in such a way that the CAD model’s geometry, shape and/or size can be changed automatically without having to manually edit the design just by changing the values in those variables as per the need. There are several CAD softwares such as Creo, CATIA, Solidworks and Fusion 360 software that allow Parametric Modelling. Fusion 360 is chosen for this study since it also offers other cutting-edge features like Generative Design and Product Design extensions. Although the Parametric Modelling approach has been existing in the CAD industry for a long time, its benefits have not been incorporated into the CAD design culture and in most scenarios, industries have been relying on the designer’s inherent creative thinking power or skills learned through training and experience to generate newer designs. This tries to show the benefits and advantages of Random Parametric modelling and how it can significantly enhance creativity in generating unique-looking models with the help of a case study/example illustrated in the next sections. A brief study of the recent developments related to the use of Deep Learning Algorithms in 3D modelling is shown in Table 1 . Green cells indicate that the authors’ work has a potential to be used in the present work, yellow cells indicate that the authors’ work is not directly relevant to present study but provides useful insights, and red cells indicate the research gap in the literature which forms the aim of the present research study. Table 1 Objective of the current study in comparison with the existing studies Model name Input dataset type Neural network model type Model objective Output data type Is the output in the form of mesh or 3D B-Rep format (Yes/No) Is the method suitable for generating random 3D models to support GD? PointNet [ 2 ] Point Clouds A benchmarked ANN with unique 3D to 1024D architecture Classification and Segmentation of Input point Clouds k-class scores or m-segments No No PointConv [ 3 ] Point Clouds Convolutional Encoder and Deconvolutional Decoder Network Classification and Segmentation k-class scores or m-class segment scores No No PointCNN [ 4 ] Point Clouds Convolutional Encoder and Deconvolutional Decoder Network with Latent Canonical Code in between Classification and Segmentation k-class scores or m-class segment scores No No ConvPoint [ 5 ] Point Clouds Convolutional Encoder and Deconvolutional Decoder Network Semantic Segmentation of the Given Point Clouds m-Segmentation scores No No SketchCleanNet [ 6 ] 2D Hand Sketches 2D Image CNN with Up-Conv and Down-Conv configuration Retrieve 3D Model from a Database based on given hand-drawn query sketch Improved 2D Image of 3D CAD Model View No No PiFUHD [ 7 ] Human Full Body 2D Images 2D Pixel to 3D Occupancy Implicit Function using Encoder-Decoder type Neural Network 3D Voxel Occupancy Based Human Body Reconstruction 3D Occupancy Function, but not a 3D Model No No BendSketch [ 8 ] 2D Hand Sketches 2D to 3D Surface remapping, mathematics involved but not Deep Learning Methods Generation of 3D Surfaces from single and multi-view hand drawn sketches Surface Boundary Representation on authors’ own software. File type not specified. Partially Yes (May be converted into a readable mesh format) No Pix2Surf [ 9 ] 2D Images of everyday objects Encoder-Decoder Network Generation of 3D Surfaces from single and multi-view hand drawn sketches Continuous Parameterised 3D Surface Patch from each view Partially Yes (May be converted into a readable mesh format) No Pix2Mesh [ 10 ] 2D Images of everyday objects, preferably those that can be modelled by deforming an ellipsoid mesh A hybrid CNN-GNN network that connects between Input 2D Images and Output 3D Meshes Generation of 3D Meshes from single view 2D Images Views of Ground Truth 3D Objects Watertight 3D Meshes Yes, the graph datatype used to represent the mesh can be converted into CAD readable formats lile STL/OBJ No Space Filling Generative Design Technique (SF-GDT) [ 11 ] Random Ranges of Parameters that can represent a parametrised CAD model This model does not use any Neural Networks to generate hybrid models Random Generation of Similar Looking CAD Models based on Parameterised Input CAD Model and filter the models based on a Cost/Loss Function Parametric CAD Models Yes, it generates a Parametric B-Rep Model Yes Fusion 360 Gym [ 12 ] Ground Truth Target Fusion 360 CAD Model (Gt) as Input Graph Neural Network + MLP to Make New/Cut/Intersect/Union operations of extruded sketch B-Rep CAD Model Reconstruction Sequence of Commands (Vector-X) consisting Sketches & Extrude Operations expressed in Fusion 360's Domain Specific Language (DSL) operated on current geometry (Gc) so that Gc can be transformed intoTarget Geometry (Gt) Yes (Fusion 360 DSL allows to convert G(X) into an B-Rep model Partially Yes (Vector-X can be randomised so that G(X) represents a randomly generated 3D B-Rep object) ShapeNet [ 13 ] N/A It is a dataset containing 2D Images and 3D Voxels of Generic 3D Objects It is a dataset and not a model It is a dataset and not a model Partially Yes (May be converted into a readable mesh format) Not a Neural Network Model, it is a Dataset of 3D Models ModelNet40 [ 14 ] N/A It is a dataset and not a model It is a dataset and not a model It is a dataset and not a model Yes, Dataset contains .OFF files Not a Neural Network Model, it is a dataset of 3D Point Clouds Present Work Random Feature Vector Artificial Neuralnetwork (Fusion-ANN) Generate Population of High-Quality 3D models 3D watertight and aesthetic B-Rep Solids (CAD Native format or STEP Format) Yes Yes Table-1 Shows that though several authors attempted to create Neural network architectures (ANNs or CNNs) to classify/segment/generate 3D data in the form of point clouds [ 2 – 5 ] , 2D Sketches [ 6 – 8 ] or Surface Meshes [ 9 , 10 ] , not much work is done to build an Artificial Neural Network or Autoencoder that can generate 3D B-Rep modelling directly from a given set of input vectors. SF-GDT [11] used random parameters to generate parametric CAD models but lacks the ability to generate unique-looking models or use AI to incorporate creativity in design. The closest development is Fusion 360 Gym [ 12 ] which is a Python based API to search design space to depict the given ground truth B-Rep model, but it lacks random generative modelling approach where a number of valid B-Rep models can be generated from a given input vector. Readymade 3D model datasets [ 13 , 14 ] are limited to only specific class of objects and do not serve for developing a general-purpose 3D modelling approach. It has been stated that there has to be an interactive method to generate designs to enhance creativity in design that can give instant feedback to the user [ 15 ] . 2. Objectives of the present work The current work aims to address this gap and the objectives are elaborated below: Create an Artificial Neural Network (ANN) model that takes in Random Latent Feature Vector / Random Parameter Values and generate design parameters sufficient to reconstruct unique and aesthetic B-Rep solids. The ANN model should have the capability to randomly generate a population of aesthetic and unique designs representing a targeted consumer product like a table lamp, a computer mouse, a joystick, a cosmetic container, or a maker-coin etc. For the purpose of demonstration, an ANN capable of generating unique ‘Maker-coins’ is developed. Output data type should be 3D aesthetic B-Rep Solids. 3. Proposed Methodology 3.1. Designing the Fusion-ANN for a 3D Printable Maker-coin Body. In the case study demonstrating the use of Neural Networks to develop B-Rep models, a simple Maker-coin is developed. Maker-coin is a coin-shaped object used as a brand identity logo in the typical 3D printing community. All the parameters used to define each maker-coin design is formulated as per the parameters and value ranges mentioned below in Table 2 . Table 2 Parameter Table with Range Values that define the geometry of each maker-coin Parameter Unit Value and Range Inscribe Circle Diameter of Base (db) mm 25 to 40 Number of Vertices of Base (nb) 3 to 10 Base Extrude Height (h) mm 3 to 5 Centre Location of Cut Circle Radial Distance (hdr) mm 4 to 20 Centre Location of Cut Circle Angular Position (cca) deg 0 to 180 Number of Vertices of Cut (nvc) 3 to 10 Inscribe Circle Diameter of Cut (dc) mm 3 to 12 Number of copies of circular array of cut (ncc) 3 to 12 3.2. Conventional Design of Maker-coin and its drawbacks In the conventional design approach, the subsequent feature development references the sketches/faces/bodies created previously, as shown in Fig. 2 . Hence any modification to sketch, face or bodies may affect all downstream features and may create modelling errors. This problem is generally not solved and designers are accustomed to having to create the models from the very scratch for each design modification demanded by the clients. 3.3. Feature Independence Approach used for creating the Maker-coin Designs. The feature independence approach (FIA) developed is elaborated below: All sketches are drawn on default origin planes or parametric planes and not on faces of previously created bodies. All sketches required to define the geometry of the body are created at once sequentially. Any sketch drawn does not refer to any of the geometric features of body like edges and also does not refer to previous sketches such as project/include. Any extrude-cut operations are be performed by first creating an extrude-add operation and then by using body-combine operation to cut features on the master body as given in Fig. 4 . Body-combine operations like bodies-union, bodies-intersect and body-cut are performed after all extrusions are completed so that there are no future operations left to be performed. Fillets if any are given at last once all features are developed and body combine operations are performed so that no geometric feature gets referenced to filleted faces or edges which may create modelling errors like missing face, missing edges etc. Fillets are prone to modelling errors when the design is changed parametrically; hence fillets will be addressed in future work. Any array patterns are only be given at the very end of the design tree so that when the pattern number changes, there will not be any features that reference faces generated by the pattern as shown in Fig. 5 . 3.4 Process for development of multiple Random Designs Microsoft Excel’s rand() function is used to get random values of the input feature vector and then the Artificial Neural Network Architecture is built using weights and biases embedded into Microsoft Excel sheet cells (with appropriate formulas). Each design set parameter values are then individually exported in .CSV format and imported into Fusion 360 as parameters using the "Parameter I/O" extension app (Fusion 360 app store) which updates the design automatically. This step is repeated 10 times to get 10 different designs and can be repeated as many as needed. Although there are numerous unique combinations possible with the range of parameters mentioned in the Table 1 , the number of possible outcomes is restricted to 10 for analysis and presentation in this work. However, the design framework can be used to generate many more possible unique designs with the proposed methodology. 3.5. Artificial Neural Network Architecture of the Fusion-ANN The architecture of the Fusion - Artificial Neural Network (Fusion-ANN) consists of the following stages: Activation Function Drop-out connectivity and rounding off Loss Function The architecture of the ANN is presented in Fig. 6 . Input layer has 8 nodes which take in random values (between 0 and 1) and Output Layer consists of another 8 nodes that produce the values of design parameters of the maker-coin. There are 2 hidden layers with 8 nodes each. Stage 1: ReLU (Rectified Linear Unit) with slope equal to 1 is used as the node activation function as it does not generate any negative values into the output layer which suits our modelling strategy. Also, negative values such as length, height and diameter are meaningless from the geometric modelling perspective which led to using ReLU. Stage 2: Our ANN does not contain fully connected node layers. Instead, only certain nodes are connected and that too only partially and remaining nodes just multiply the previous corresponding node value with the weight and pass onto the next layer. This is done to make sure that certain input feature values stay independent of other features. Round-off function is applied at the output layer to round off the decimal values into integral values. Stage 3: Since the aim of this work is to produce aesthetic models, there is no mathematical way to quantify how aesthetic the generated model is and so the aesthetic score has to be assigned manually (by looking at each generated 3D model). The aesthetic scores are assigned as per the Table 3 . Table 3 Guidelines followed for manual assignment of Aesthetic Score to the generated maker-coin model. If the model has: Score Assigned Shape resembling a closed polygon (Score 1) 4 Peripheral cut operations did not split the body into fragments (Score 2) 3 Model did not contain any sharp pockets/edges which enable it for manufacturing at a lower cost (Score 3) 3 Total (Score 1 + Score 2 + Score 3) 10 Hence at this stage of the present work, any loss functions to train and update the weights is not used. Nevertheless, the generated models are aesthetic enough (presented in results table) without even training the ANN which highlights the suitability of this method in creating new designs in scenarios where 3D model datasets do not exist for the type/class of object in need. 4. Results and Discussion The below table shows the randomly generated input vector values that are fed into our ANN. MD1 denotes maker-coin design 1, MD 2 denotes maker-coin design 2 and so on. Table 4 Input Feature Vector Table initialised randomly for each design Input Feature Node Number MD1 MD2 MD3 MD4 MD5 MD6 MD7 MD8 MD9 MD10 1 0.849 0.136 0.301 0.244 0.079 0.961 0.144 0.579 0.161 0.437 2 0.505 0.738 0.752 0.126 0.167 0.344 0.580 0.260 0.902 0.857 3 0.556 0.999 0.568 0.540 0.378 0.053 0.324 0.786 0.691 0.262 4 0.554 0.677 0.869 0.439 0.065 0.462 0.009 0.657 0.208 0.336 5 0.757 0.368 0.819 0.415 0.771 0.904 0.865 0.420 0.31 0.255 6 0.693 0.292 0.007 0.981 0.935 0.283 0.680 0.706 0.776 0.215 7 0.049 0.606 0.730 0.582 0.975 0.531 0.262 0.909 0.079 0.627 8 0.246 0.351 0.340 0.394 0.524 0.211 0.648 0.676 0.79 0.705 Below table shows the output parameter values for each candidate designs generated by the Neural Network. Below table shows the output parameter values for each candidate designs generated by the Neural Network. Table 5 Candidate Maker-coin Design Parameters (MD-1 to MD-10) generated by the Neural Network Parameter MD-1 MD-2 MD-3 MD-4 MD-5 MD-6 MD-7 MD-8 MD-9 MD-10 db 38 27 30 29 26 39 27 34 27 32 hdr 14 11 13 7 5 13 8 11 10 13 dc 8 12 8 7 6 3 5 10 9 5 nb 7 8 9 6 3 6 3 8 4 5 ncc 8 7 11 6 5 9 5 7 3 4 nvc 8 5 3 10 10 5 8 8 8 5 cca 9 109 131 105 176 96 47 164 14 113 h 4 4 4 4 5 4 5 5 5 5 We are presenting the final designs reconstructed automatically by the Parameter I/O extension in Fusion 360 in the below table 6. Out of the 10 designs generated randomly, 3 designs qualified for presenting as design alternatives before the client (with scores 8 or above) with MD-3 & MD-6 scoring 10 points and standing out as the most aesthetic looking maker-coins, MD-10 scoring 9 points, and MD-9 scoring 8 points. Three designs (MD-2, MD-5, MD-8) have less than 5 points and failed to qualify for presenting before the client. MD-4 and MD-7 have a potential to be good candidates if their feature vector is tweaked a bit. Overall, the developed Neural network when fed into the Fusion 360 as parameters yielded some stunning and unique designs that have a potential value to the client (the maker-coin can be 3D printed and showcased to reflect the brand identify) and additionally, the Neural Network is easy to compute which does not involve any computationally intensive training. This methodology to encode design features into a latent feature vector and then using that to reconstruct as many designs as the user wants is the highlight of this research as shown in Fig. 7 . Table-6: Images of the Maker-coin Designs MD-1 to MD-10 reconstructed using the parameters generated by the Neural Network and their corresponding aesthetic score Design Polygon Score De-Fragmentation Score Ease of Manufacturing Score Total Aesthetic Score (10-point scale) MD-1 4 3 0 7 MD-2 4 0 0 4 MD-3 4 3 3 10 MD-4 4 2 1 7 MD-5 2 0 0 2 MD-6 4 3 3 10 MD-7 4 3 0 7 MD-8 3 0 0 3 MD-9 2 3 3 8 MD-10 4 3 2 9 5. Conclusions From the results and discussions, it is successfully demonstrated that the proposed Neural Network combined with parametric reconstruction modelling in Fusion 360 is suitable for creating complex creative designs with the help of the maker-coin example. The limitations of conventional modelling approach are addressed and how the proposed Feature-Independent approach can help reduce the modelling errors are highlighted. Some of the applications of the proposed methodology are: Creation of large geometric, form and design variations for presenting concept alternatives in scenarios where creativity, randomness and uniqueness play a role in impressing the client/customer. Creation of large CAD model datasets from a few hundred to a few thousand unique but similar designs for 3D Deep Learning use cases. Creation of a large set of Digital CAD models as a collection of Non-Fungible Tokens (NFTs). Use genetic algorithms to further refine the design search space and generate better designs by feeding the input feature vector as the DNA and evolving the population in each generation. Results of the present methods suggests to follow the random search methods as they do not have mathematical loss function to train itself to produce more aesthetically looking models for best possible designs. References “The Next Wave of Intelligent Design Automation”. Harvard Business Review, Sponsored by Autodesk. Issue Date: June 2018. URL: https://hbr.org/sponsored/2018/06/the-next-wave-of-intelligent-design-automation?autocomplete=true [Date Accessed: 09-Oct-2022] Charles R. Qi, Hao Su, Kaichun Mo and Leonidas J. Guibas. “PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation”. Conference on Computer Vision and Pattern Recognition (CVPR) 2017. https://doi.org/10.48550/arXiv.1612.00593 Wenxuan Wu, Zhongang Qi, Li Fuxin.“PointConv: Deep Convolutional Networks on 3D Point Clouds”. Conference on Computer Vision and Pattern Recognition (CVPR) 2019. https://doi.org/10.48550/arXiv.1811.07246 Yangyan Li, Rui Bu, Mingchao Sun, Wei Wu, Xinhan Di, Baoquan Chen. “PointCNN: Convolution On X -Transformed Points”. Conference on Neuralnetwork Information Processing Systems 2018. https://doi.org/10.48550/arXiv.1801.07791 Boulch A. “ConvPoint: Continuous Convolutions for Point Cloud Processing”. 2019. https://doi.org/10.48550/ARXIV.1904.02375 Manda B, Kendre P, Dey S, Muthuganapathy R. SketchCleanNet – “A deep learning approach to the enhancement and correction of query sketches for a 3D CAD model retrieval system”. 2022. https://doi.org/10.48550/ARXIV.2207.00732 Saito S, Simon T, Saragih J, Joo H. “PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization”. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2020. https://doi.org/10.1109/cvpr42600.2020.00016 Changjian Li, Hao Pan, Yang Liu, Xin Tong, Alla Sheffer, and Wenping Wang. 2017. “BendSketch: Modeling Freeform Surfaces Through 2D Sketching”. ACM Trans. Graph. 36, 4, Article 125 (July 2017), 14 pages. http://dx.doi.org/10.1145/3072959.3073632 Lei J, Sridhar S, Guerrero P, Sung M, Mitra N, Guibas LJ. “Pix2Surf: Learning Parametric 3D Surface Models of Objects from Images”. 2020. https://doi.org/10.48550/ARXIV.2008.07760 Wang N, Zhang Y, Li Z, Fu Y, Liu W, Jiang Y-G. “Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images”. ArXiv 2018. https://doi.org/10.48550/ARXIV.1804.01654 Khan S, Awan MJ. “A generative design technique for exploring shape variations”. Advanced Engineering Informatics 2018; 38:712–24. https://doi.org/10.1016/j.aei.2018.10.005 Fusion 360 GYM, a Python API for ML models to be operated on Fusion 360 Gallery https://github.com/AutodeskAILab/Fusion360GalleryDataset/tree/master/tools/fusion360gym [Dataset] ShapeNet -https://shapenet.org/ [Dataset] ModelNet40 - https://paperswithcode.com/sota/3d-point-cloud-classification-on-modelnet40 Li Y, Wang J, Li X, Zhao W. “Design creativity in product innovation”. Int J Adv Manuf Technol 2006; 33:213–22. https://doi.org/10.1007/s00170-006-0457-y Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3126805","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":214910391,"identity":"416e25fd-1828-4866-b24e-41439d901106","order_by":0,"name":"Nikhil Chaitanya Angajala","email":"","orcid":"","institution":"GITAM University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Nikhil","middleName":"Chaitanya","lastName":"Angajala","suffix":""},{"id":214910392,"identity":"b3897380-bd52-4391-9fd3-6eed826d0462","order_by":1,"name":"Srinivasa Prasad Balla","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYBACAwQz+QCQkJAhRUtaAkgLDylacsBswlrM+c8eYC7cY2fP357z+dWNGgseBvbDRzfg02I5Iy+Becaz5MQZZ95us845BnQYT1raDbwOu8FjwMxzgDmB4UbuNuMcNqAWCR4z/FrOnwFpqbeXv5HzzDjnHzFaDuSAtBxm3HAjh/lxbhsRWkB+OTzjwPHEjWeemTHn9knwsBHyCzDEDj4uOFBtL3c8+fHnnG91cvzsh4/h1QKKiMNQFpsEmMSvHKKFGcpi/kBY9SgYBaNgFIxEAAAoqEfKmLF+ugAAAABJRU5ErkJggg==","orcid":"","institution":"GITAM University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Srinivasa","middleName":"Prasad","lastName":"Balla","suffix":""},{"id":214910393,"identity":"a12bf6cc-081e-4a8f-ac91-c85c767cfcdf","order_by":2,"name":"Rajeswara Reddy Resapu","email":"","orcid":"","institution":"GITAM University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Rajeswara","middleName":"Reddy","lastName":"Resapu","suffix":""}],"badges":[],"createdAt":"2023-06-30 04:44:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3126805/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3126805/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":39639840,"identity":"220a9c17-528b-4e76-a4c4-6e49c0fea828","added_by":"auto","created_at":"2023-07-06 14:15:03","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":64406,"visible":true,"origin":"","legend":"\u003cp\u003eScope of the proposed work\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/93504ee6521cbf65e015c2b0.png"},{"id":39639835,"identity":"e3967d0e-df82-4cfe-a877-247f662ce33c","added_by":"auto","created_at":"2023-07-06 14:15:03","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":335048,"visible":true,"origin":"","legend":"\u003cp\u003eDesign History Tree of Conventional Design Approach\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/403ae52023a7e67142b3fd9d.jpg"},{"id":39639837,"identity":"044da54a-efc2-4e22-9b94-e6cb38c960bf","added_by":"auto","created_at":"2023-07-06 14:15:03","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":97054,"visible":true,"origin":"","legend":"\u003cp\u003eMaker-coin designed using a conventional design approach.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/b054846ddb6bd73e300b0505.jpg"},{"id":39639838,"identity":"00b142e2-2ca4-4aa3-9dff-967dd3b1f62e","added_by":"auto","created_at":"2023-07-06 14:15:03","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":304078,"visible":true,"origin":"","legend":"\u003cp\u003eDesign History Tree of Feature-Independent Design Approach.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/536ef1460e62b76fbef9f4ec.jpg"},{"id":39640960,"identity":"53dda20c-6180-4848-96d5-d439c4729fae","added_by":"auto","created_at":"2023-07-06 14:23:03","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":79868,"visible":true,"origin":"","legend":"\u003cp\u003eMaker-coin designed using the proposed Fusion-ANN\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/3690e08a36c26f2e869fd906.jpg"},{"id":39639841,"identity":"a06687a9-db7a-4427-9bfb-40b286ee6255","added_by":"auto","created_at":"2023-07-06 14:15:03","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":472093,"visible":true,"origin":"","legend":"\u003cp\u003eArchitecture of the ANN used for the maker-coin design\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/9e584af634a83fdf9521d501.jpg"},{"id":39639836,"identity":"08a17614-8085-46e0-b386-5d7974f2ae6c","added_by":"auto","created_at":"2023-07-06 14:15:03","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":138186,"visible":true,"origin":"","legend":"\u003cp\u003eDesign outcomes MD-1 to MD-10\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/8a4f66b94f5d4a4a19ff2f4e.png"},{"id":39699499,"identity":"59669158-37dd-442b-aa7c-02e80cd33dc4","added_by":"auto","created_at":"2023-07-07 13:59:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":949685,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3126805/v1/29b48eeb-e6fe-4dfa-9964-781414b4b502.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Analysis on Novel Feature-Independent CAD Design/Modelling Approach Combined with a Neural Network for Random Parametric 3D Boundary Representation Modelling","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eA recent (2018) Harvard Business Review article \u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e stated that although CAD (Computer Aided Design) was introduced in late 20th century to aid designers and engineers in creation of cad models, they significantly lack the \u0026ldquo;computer aiding\u0026rdquo; element and merely support only the documenting framework of what the designer/engineer is imagining in his head and the article suggested and predicted that Generative Design (GD) methods will be the next big wave in transforming the design approach across various engineering domains. As far as the GD modules available in present-day CAD softwares like Fusion 360, the design alternatives are based on the specified constraints like (i) Geometric Constraints such as obstacle geometries and geometric regions to retain, (ii) Structural Constraints specified in terms of load, factor of safety and mass, (iii) Material Constraints like Tensile Strength and (iv) Manufacturing Constraints like process restrictions that are specific to the manufacturing process like Machining, Die Casting and Additive Manufacturing. Although GD is a big step towards utilising Artificial Intelligence (AI) to create design solutions, it lacks the ability to generate aesthetic models and also lacks creativity in terms of possible design solutions that are specific to consumer product design such as design of outer body of electronic gadgets, body panels of automobiles, surfaces of cosmetic packaging etc. Hence these gaps form as the basis and motivation for the present study.\u003c/p\u003e \u003cp\u003eParametric Modelling refers to the methodology of defining the design variables (or otherwise known as parameters and hence the name \u0026lsquo;Parametric\u0026rsquo; modelling) that store the values of dimensions and constraints besides defining other mathematical relations that govern the behaviour of the model that make up the CAD model in such a way that the CAD model\u0026rsquo;s geometry, shape and/or size can be changed automatically without having to manually edit the design just by changing the values in those variables as per the need. There are several CAD softwares such as Creo, CATIA, Solidworks and Fusion 360 software that allow Parametric Modelling. Fusion 360 is chosen for this study since it also offers other cutting-edge features like Generative Design and Product Design extensions. Although the Parametric Modelling approach has been existing in the CAD industry for a long time, its benefits have not been incorporated into the CAD design culture and in most scenarios, industries have been relying on the designer\u0026rsquo;s inherent creative thinking power or skills learned through training and experience to generate newer designs. This tries to show the benefits and advantages of Random Parametric modelling and how it can significantly enhance creativity in generating unique-looking models with the help of a case study/example illustrated in the next sections.\u003c/p\u003e \u003cp\u003eA brief study of the recent developments related to the use of Deep Learning Algorithms in 3D modelling is shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Green cells indicate that the authors\u0026rsquo; work has a potential to be used in the present work, yellow cells indicate that the authors\u0026rsquo; work is not directly relevant to present study but provides useful insights, and red cells indicate the research gap in the literature which forms the aim of the present research study.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eObjective of the current study in comparison with the existing studies\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel name\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInput dataset type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNeural network model type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModel objective\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOutput data type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eIs the output in the form of mesh or 3D B-Rep format (Yes/No)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eIs the method suitable for generating random 3D models to support GD?\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePointNet \u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoint Clouds\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eA benchmarked ANN with unique 3D to 1024D architecture\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClassification and Segmentation of Input point Clouds\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ek-class scores or m-segments\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePointConv \u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoint Clouds\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eConvolutional Encoder and Deconvolutional Decoder Network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClassification and Segmentation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ek-class scores or m-class segment scores\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePointCNN \u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoint Clouds\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eConvolutional Encoder and Deconvolutional Decoder Network with Latent Canonical Code in between\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClassification and Segmentation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ek-class scores or m-class segment scores\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConvPoint \u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoint Clouds\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eConvolutional Encoder and Deconvolutional Decoder Network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSemantic Segmentation of the Given Point Clouds\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em-Segmentation scores\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSketchCleanNet \u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2D Hand Sketches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2D Image CNN with Up-Conv and Down-Conv configuration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRetrieve 3D Model from a Database based on given hand-drawn query sketch\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eImproved 2D Image of 3D CAD Model View\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePiFUHD \u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHuman Full Body 2D Images\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2D Pixel to 3D Occupancy Implicit Function using Encoder-Decoder type Neural Network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3D Voxel Occupancy Based Human Body Reconstruction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3D Occupancy Function, but not a 3D Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBendSketch \u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2D Hand Sketches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2D to 3D Surface remapping, mathematics involved but not Deep Learning Methods\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGeneration of 3D Surfaces from single and multi-view hand drawn sketches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSurface Boundary Representation on authors\u0026rsquo; own software. File type not specified.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePartially Yes (May be converted into a readable mesh format)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePix2Surf \u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2D Images of everyday objects\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEncoder-Decoder Network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGeneration of 3D Surfaces from single and multi-view hand drawn sketches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eContinuous Parameterised 3D Surface Patch from each view\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePartially Yes (May be converted into a readable mesh format)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePix2Mesh \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2D Images of everyday objects, preferably those that can be modelled by deforming an ellipsoid mesh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eA hybrid CNN-GNN network that connects between Input 2D Images and Output 3D Meshes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGeneration of 3D Meshes from single view 2D Images Views of Ground Truth 3D Objects\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eWatertight 3D Meshes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes, the graph datatype used to represent the mesh can be converted into CAD readable formats lile STL/OBJ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpace Filling Generative Design Technique (SF-GDT) \u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRandom Ranges of Parameters that can represent a parametrised CAD model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThis model does not use any Neural Networks to generate hybrid models\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRandom Generation of Similar Looking CAD Models based on Parameterised Input CAD Model and filter the models based on a Cost/Loss Function\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eParametric CAD Models\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes, it generates a Parametric B-Rep Model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFusion 360 Gym \u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGround Truth Target Fusion 360 CAD Model (Gt) as Input\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGraph Neural Network\u0026thinsp;+\u0026thinsp;MLP to Make New/Cut/Intersect/Union operations of extruded sketch\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eB-Rep CAD Model Reconstruction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSequence of Commands (Vector-X) consisting Sketches \u0026amp; Extrude Operations expressed in Fusion 360's Domain Specific Language (DSL) operated on current geometry (Gc) so that Gc can be transformed intoTarget Geometry (Gt)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes (Fusion 360 DSL allows to convert G(X) into an B-Rep model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePartially Yes (Vector-X can be randomised so that G(X) represents a randomly generated 3D B-Rep object)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eShapeNet \u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIt is a dataset containing 2D Images and 3D Voxels of Generic 3D Objects\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIt is a dataset and not a model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIt is a dataset and not a model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePartially Yes (May be converted into a readable mesh format)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNot a Neural Network Model, it is a Dataset of 3D Models\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModelNet40 \u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIt is a dataset and not a model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIt is a dataset and not a model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIt is a dataset and not a model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes, Dataset contains .OFF files\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNot a Neural Network Model, it is a dataset of 3D Point Clouds\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent Work\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRandom Feature Vector\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eArtificial Neuralnetwork (Fusion-ANN)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGenerate Population of High-Quality 3D models\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3D watertight and aesthetic B-Rep Solids (CAD Native format or STEP Format)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable-1 Shows that though several authors attempted to create Neural network architectures (ANNs or CNNs) to classify/segment/generate 3D data in the form of point clouds \u003csup\u003e[\u003cspan additionalcitationids=\"CR3 CR4\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e, 2D Sketches \u003csup\u003e[\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e or Surface Meshes \u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e, not much work is done to build an Artificial Neural Network or Autoencoder that can generate 3D B-Rep modelling directly from a given set of input vectors. SF-GDT [11] used random parameters to generate parametric CAD models but lacks the ability to generate unique-looking models or use AI to incorporate creativity in design. The closest development is Fusion 360 Gym \u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e which is a Python based API to search design space to depict the given ground truth B-Rep model, but it lacks random generative modelling approach where a number of valid B-Rep models can be generated from a given input vector. Readymade 3D model datasets \u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e are limited to only specific class of objects and do not serve for developing a general-purpose 3D modelling approach. It has been stated that there has to be an interactive method to generate designs to enhance creativity in design that can give instant feedback to the user \u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e"},{"header":"2. Objectives of the present work","content":"\u003cp\u003eThe current work aims to address this gap and the objectives are elaborated below:\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eCreate an Artificial Neural Network (ANN) model that takes in Random Latent Feature Vector / Random Parameter Values and generate design parameters sufficient to reconstruct unique and aesthetic B-Rep solids.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eThe ANN model should have the capability to randomly generate a population of aesthetic and unique designs representing a targeted consumer product like a table lamp, a computer mouse, a joystick, a cosmetic container, or a maker-coin etc. For the purpose of demonstration, an ANN capable of generating unique ‘Maker-coins’ is developed.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eOutput data type should be 3D aesthetic B-Rep Solids.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003cp\u003e\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003cp\u003e\u003c/p\u003e \u003c/div\u003e "},{"header":"3. Proposed Methodology","content":"\u003ch2\u003e3.1. Designing the Fusion-ANN for a 3D Printable Maker-coin Body.\u003c/h2\u003e\n\u003cp\u003eIn the case study demonstrating the use of Neural Networks to develop B-Rep models, a simple Maker-coin is developed. Maker-coin is a coin-shaped object used as a brand identity logo in the typical 3D printing community. All the parameters used to define each maker-coin design is formulated as per the parameters and value ranges mentioned below in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab2\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eParameter Table with Range Values that define the geometry of each maker-coin\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\u003ccolgroup\u003e\u003c/colgroup\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eParameter\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eUnit\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eValue and Range\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eInscribe Circle Diameter of Base (db)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003emm\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e25 to 40\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNumber of Vertices of Base (nb)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3 to 10\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBase Extrude Height (h)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003emm\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3 to 5\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCentre Location of Cut Circle Radial Distance (hdr)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003emm\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e4 to 20\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCentre Location of Cut Circle Angular Position (cca)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003edeg\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e0 to 180\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNumber of Vertices of Cut (nvc)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3 to 10\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eInscribe Circle Diameter of Cut (dc)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003emm\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3 to 12\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNumber of copies of circular array of cut (ncc)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3 to 12\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003ch2\u003e3.2. Conventional Design of Maker-coin and its drawbacks\u003c/h2\u003e\n\u003cp\u003eIn the conventional design approach, the subsequent feature development references the sketches/faces/bodies created previously, as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. Hence any modification to sketch, face or bodies may affect all downstream features and may create modelling errors. This problem is generally not solved and designers are accustomed to having to create the models from the very scratch for each design modification demanded by the clients.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003e3.3. Feature Independence Approach used for creating the Maker-coin Designs.\u003c/h2\u003e\n\u003cp\u003eThe feature independence approach (FIA) developed is elaborated below:\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eAll sketches are drawn on default origin planes or parametric planes and not on faces of previously created bodies.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAll sketches required to define the geometry of the body are created at once sequentially. Any sketch drawn does not refer to any of the geometric features of body like edges and also does not refer to previous sketches such as project/include.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAny extrude-cut operations are be performed by first creating an extrude-add operation and then by using body-combine operation to cut features on the master body as given in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eBody-combine operations like bodies-union, bodies-intersect and body-cut are performed after all extrusions are completed so that there are no future operations left to be performed.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003eFillets if any are given at last once all features are developed and body combine operations are performed so that no geometric feature gets referenced to filleted faces or edges which may create modelling errors like missing face, missing edges etc. Fillets are prone to modelling errors when the design is changed parametrically; hence fillets will be addressed in future work.\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAny array patterns are only be given at the very end of the design tree so that when the pattern number changes, there will not be any features that reference faces generated by the pattern as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003e3.4 Process for development of multiple Random Designs\u003c/h2\u003e\n\u003cp\u003eMicrosoft Excel\u0026rsquo;s rand() function is used to get random values of the input feature vector and then the Artificial Neural Network Architecture is built using weights and biases embedded into Microsoft Excel sheet cells (with appropriate formulas). Each design set parameter values are then individually exported in .CSV format and imported into Fusion 360 as parameters using the \"Parameter I/O\" extension app (Fusion 360 app store) which updates the design automatically. This step is repeated 10 times to get 10 different designs and can be repeated as many as needed. Although there are numerous unique combinations possible with the range of parameters mentioned in the Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e, the number of possible outcomes is restricted to 10 for analysis and presentation in this work. However, the design framework can be used to generate many more possible unique designs with the proposed methodology.\u003c/p\u003e\n\u003ch2\u003e3.5. Artificial Neural Network Architecture of the Fusion-ANN\u003c/h2\u003e\n\u003cp\u003eThe architecture of the Fusion - Artificial Neural Network (Fusion-ANN) consists of the following stages:\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eActivation Function\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eDrop-out connectivity and rounding off\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eLoss Function\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe architecture of the ANN is presented in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e. Input layer has 8 nodes which take in random values (between 0 and 1) and Output Layer consists of another 8 nodes that produce the values of design parameters of the maker-coin. There are 2 hidden layers with 8 nodes each.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eStage 1: ReLU (Rectified Linear Unit) with slope equal to 1 is used as the node activation function as it does not generate any negative values into the output layer which suits our modelling strategy. Also, negative values such as length, height and diameter are meaningless from the geometric modelling perspective which led to using ReLU.\u003c/p\u003e\n\u003cp\u003eStage 2: Our ANN does not contain fully connected node layers. Instead, only certain nodes are connected and that too only partially and remaining nodes just multiply the previous corresponding node value with the weight and pass onto the next layer. This is done to make sure that certain input feature values stay independent of other features. Round-off function is applied at the output layer to round off the decimal values into integral values.\u003c/p\u003e\n\u003cp\u003eStage 3: Since the aim of this work is to produce aesthetic models, there is no mathematical way to quantify how aesthetic the generated model is and so the aesthetic score has to be assigned manually (by looking at each generated 3D model). The aesthetic scores are assigned as per the Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab3\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eGuidelines followed for manual assignment of Aesthetic Score to the generated maker-coin model.\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\u003ccolgroup\u003e\u003c/colgroup\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eIf the model has:\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eScore Assigned\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eShape resembling a closed polygon (Score 1)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePeripheral cut operations did not split the body into fragments (Score 2)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eModel did not contain any sharp pockets/edges which enable it for manufacturing at a lower cost (Score 3)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eTotal (Score 1\u0026thinsp;+\u0026thinsp;Score 2\u0026thinsp;+\u0026thinsp;Score 3)\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u003cstrong\u003e10\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eHence at this stage of the present work, any loss functions to train and update the weights is not used. Nevertheless, the generated models are aesthetic enough (presented in results table) without even training the ANN which highlights the suitability of this method in creating new designs in scenarios where 3D model datasets do not exist for the type/class of object in need.\u003c/p\u003e"},{"header":"4. Results and Discussion","content":"\u003cp\u003eThe below table shows the randomly generated input vector values that are fed into our ANN. MD1 denotes maker-coin design 1, MD 2 denotes maker-coin design 2 and so on.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab4\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eInput Feature Vector Table initialised randomly for each design\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eInput Feature Node Number\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD1\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD2\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD3\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD4\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD5\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD6\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD7\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD8\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD9\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD10\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e1\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.849\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.136\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.301\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.244\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.079\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.961\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.144\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.579\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.161\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.437\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.505\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.738\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.752\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.126\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.167\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.344\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.580\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.260\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.902\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.857\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.556\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.999\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.568\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.540\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.378\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.053\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.324\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.786\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.691\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.262\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.554\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.677\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.869\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.439\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.065\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.462\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.009\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.657\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.208\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.336\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.757\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.368\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.819\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.415\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.771\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.904\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.865\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.420\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.31\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.255\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.693\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.292\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.007\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.981\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.935\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.283\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.680\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.706\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.776\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.215\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.049\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.606\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.730\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.582\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.975\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.531\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.262\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.909\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.079\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.627\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.246\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.351\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.340\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.394\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.524\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.211\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.648\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.676\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.79\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.705\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003ctfoot\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"11\"\u003eBelow table shows the output parameter values for each candidate designs generated by the Neural Network.\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tfoot\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eBelow table shows the output parameter values for each candidate designs generated by the Neural Network.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003cdiv class=\"colspec\" align=\"char\"\u003e\u0026nbsp;\u003c/div\u003e\n\u003ctable id=\"Tab5\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eCandidate Maker-coin Design Parameters (MD-1 to MD-10) generated by the Neural Network\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eParameter\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-1\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-2\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-3\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-4\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-5\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-6\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-7\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-8\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-9\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMD-10\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003edb\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e38\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e27\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e30\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e29\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e26\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e39\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e27\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e34\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e27\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e32\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ehdr\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e14\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e11\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e13\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e13\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e11\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e10\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e13\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003edc\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e12\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e10\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e9\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003enb\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e9\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003encc\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e11\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e9\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003envc\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e10\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e10\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ecca\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e9\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e109\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e131\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e105\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e176\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e96\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e47\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e164\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e14\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e113\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eh\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e5\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eWe are presenting the final designs reconstructed automatically by the Parameter I/O extension in Fusion 360 in the below table 6.\u003c/p\u003e\n\u003cp\u003eOut of the 10 designs generated randomly, 3 designs qualified for presenting as design alternatives before the client (with scores 8 or above) with MD-3 \u0026amp; MD-6 scoring 10 points and standing out as the most aesthetic looking maker-coins, MD-10 scoring 9 points, and MD-9 scoring 8 points. Three designs (MD-2, MD-5, MD-8) have less than 5 points and failed to qualify for presenting before the client. MD-4 and MD-7 have a potential to be good candidates if their feature vector is tweaked a bit. Overall, the developed Neural network when fed into the Fusion 360 as parameters yielded some stunning and unique designs that have a potential value to the client (the maker-coin can be 3D printed and showcased to reflect the brand identify) and additionally, the Neural Network is easy to compute which does not involve any computationally intensive training. This methodology to encode design features into a latent feature vector and then using that to reconstruct as many designs as the user wants is the highlight of this research as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eTable-6: Images of the Maker-coin Designs MD-1 to MD-10 reconstructed using the parameters generated by the Neural Network and their corresponding aesthetic score\u003c/p\u003e\n\u003ctable border=\"1\" width=\"100%\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003e\u003cstrong\u003eDesign\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e\u003cstrong\u003ePolygon Score\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e\u003cstrong\u003eDe-Fragmentation Score\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e\u003cstrong\u003eEase of Manufacturing Score\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e\u003cstrong\u003eTotal Aesthetic Score \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(10-point scale)\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-1\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e\u0026nbsp;3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e\u0026nbsp;0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e\u0026nbsp;7\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-2\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e\u003cstrong\u003e10\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e2\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e1\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-5\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e2\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e2\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e\u003cstrong\u003e10\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-7\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e7\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-8\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e0\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-9\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e2\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e8\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"10%\"\u003e\n\u003cp\u003eMD-10\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"11%\"\u003e\n\u003cp\u003e4\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"22%\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"30%\"\u003e\n\u003cp\u003e2\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"25%\"\u003e\n\u003cp\u003e9\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eFrom the results and discussions, it is successfully demonstrated that the proposed Neural Network combined with parametric reconstruction modelling in Fusion 360 is suitable for creating complex creative designs with the help of the maker-coin example. The limitations of conventional modelling approach are addressed and how the proposed Feature-Independent approach can help reduce the modelling errors are highlighted. Some of the applications of the proposed methodology are:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eCreation of large geometric, form and design variations for presenting concept alternatives in scenarios where creativity, randomness and uniqueness play a role in impressing the client/customer.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eCreation of large CAD model datasets from a few hundred to a few thousand unique but similar designs for 3D Deep Learning use cases.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eCreation of a large set of Digital CAD models as a collection of Non-Fungible Tokens (NFTs).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eUse genetic algorithms to further refine the design search space and generate better designs by feeding the input feature vector as the DNA and evolving the population in each generation.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eResults of the present methods suggests to follow the random search methods as they do not have mathematical loss function to train itself to produce more aesthetically looking models for best possible designs.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e\u0026ldquo;The Next Wave of Intelligent Design Automation\u0026rdquo;. Harvard Business Review, Sponsored by Autodesk. Issue Date: June 2018. URL: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://hbr.org/sponsored/2018/06/the-next-wave-of-intelligent-design-automation?autocomplete=true\u003c/span\u003e\u003cspan address=\"https://hbr.org/sponsored/2018/06/the-next-wave-of-intelligent-design-automation?autocomplete=true\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e [Date Accessed: 09-Oct-2022]\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCharles R. Qi, Hao Su, Kaichun Mo and Leonidas J. Guibas. \u0026ldquo;PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation\u0026rdquo;. Conference on Computer Vision and Pattern Recognition (CVPR) 2017. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.1612.00593\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.1612.00593\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWenxuan Wu, Zhongang Qi, Li Fuxin.\u0026ldquo;PointConv: Deep Convolutional Networks on 3D Point Clouds\u0026rdquo;. Conference on Computer Vision and Pattern Recognition (CVPR) 2019. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.1811.07246\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.1811.07246\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYangyan Li, Rui Bu, Mingchao Sun, Wei Wu, Xinhan Di, Baoquan Chen. \u0026ldquo;PointCNN: Convolution On X -Transformed Points\u0026rdquo;. Conference on Neuralnetwork Information Processing Systems 2018. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.1801.07791\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.1801.07791\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoulch A. \u0026ldquo;ConvPoint: Continuous Convolutions for Point Cloud Processing\u0026rdquo;. 2019. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.1904.02375\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.1904.02375\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eManda B, Kendre P, Dey S, Muthuganapathy R. SketchCleanNet \u0026ndash; \u0026ldquo;A deep learning approach to the enhancement and correction of query sketches for a 3D CAD model retrieval system\u0026rdquo;. 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2207.00732\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2207.00732\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaito S, Simon T, Saragih J, Joo H. \u0026ldquo;PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization\u0026rdquo;. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/cvpr42600.2020.00016\u003c/span\u003e\u003cspan address=\"10.1109/cvpr42600.2020.00016\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChangjian Li, Hao Pan, Yang Liu, Xin Tong, Alla Sheffer, and Wenping Wang. 2017. \u0026ldquo;BendSketch: Modeling Freeform Surfaces Through 2D Sketching\u0026rdquo;. ACM Trans. Graph. 36, 4, Article 125 (July 2017), 14 pages. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://dx.doi.org/10.1145/3072959.3073632\u003c/span\u003e\u003cspan address=\"10.1145/3072959.3073632\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLei J, Sridhar S, Guerrero P, Sung M, Mitra N, Guibas LJ. \u0026ldquo;Pix2Surf: Learning Parametric 3D Surface Models of Objects from Images\u0026rdquo;. 2020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2008.07760\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2008.07760\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang N, Zhang Y, Li Z, Fu Y, Liu W, Jiang Y-G. \u0026ldquo;Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images\u0026rdquo;. ArXiv 2018. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.1804.01654\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.1804.01654\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhan S, Awan MJ. \u0026ldquo;A generative design technique for exploring shape variations\u0026rdquo;. Advanced Engineering Informatics 2018; 38:712\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.aei.2018.10.005\u003c/span\u003e\u003cspan address=\"10.1016/j.aei.2018.10.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFusion 360 GYM, a Python API for ML models to be operated on Fusion 360 Gallery \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/AutodeskAILab/Fusion360GalleryDataset/tree/master/tools/fusion360gym\u003c/span\u003e\u003cspan address=\"https://github.com/AutodeskAILab/Fusion360GalleryDataset/tree/master/tools/fusion360gym\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e[Dataset] ShapeNet -https://shapenet.org/\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e[Dataset] ModelNet40 - \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://paperswithcode.com/sota/3d-point-cloud-classification-on-modelnet40\u003c/span\u003e\u003cspan address=\"https://paperswithcode.com/sota/3d-point-cloud-classification-on-modelnet40\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, Wang J, Li X, Zhao W. \u0026ldquo;Design creativity in product innovation\u0026rdquo;. Int J Adv Manuf Technol 2006; 33:213\u0026ndash;22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00170-006-0457-y\u003c/span\u003e\u003cspan address=\"10.1007/s00170-006-0457-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Parametric Modelling, Fusion 360, Deep Learning, Generative Design, Product Design, Neural Network","lastPublishedDoi":"10.21203/rs.3.rs-3126805/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3126805/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn the era of competitive digital innovation, product manufacturing companies need rapid customisation and ability to create uniqueness in new product development to stay competitive in the consumer market. Till recent times, this requirement heavily depended on CAD designer\u0026rsquo;s ability and experience to produce creative designs. This paper presents a novel feature-independent CAD design/modelling approach combined with a neural network that enables the creation of random parametric 3D CAD design variants using the Boundary Representation (B-Rep) method. This method is rapid and thus offers about 10 to 100 concept alternatives to the client in 10 to 30 minutes. Additionally, the paper also highlights the suitability of the proposed Neural Network method in creating 3D deep learning datasets to train generative design models like 3D GANs to further enhance 3D designs specifically targeted in product design problems.\u003c/p\u003e","manuscriptTitle":"Analysis on Novel Feature-Independent CAD Design/Modelling Approach Combined with a Neural Network for Random Parametric 3D Boundary Representation Modelling","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-07-06 14:14:58","doi":"10.21203/rs.3.rs-3126805/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"55a66fd3-8f9b-46ec-811c-60d7aebcc878","owner":[],"postedDate":"July 6th, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2023-07-12T12:29:31+00:00","versionOfRecord":[],"versionCreatedAt":"2023-07-06 14:14:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3126805","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3126805","identity":"rs-3126805","version":["v1"]},"buildId":"FbvkV6FR0MCFSLy54lSbu","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}