REPRESENTATION LEARNING OF MODEL DATA

    公开(公告)号:US20250077874A1

    公开(公告)日:2025-03-06

    申请号:US18457854

    申请日:2023-08-29

    Applicant: Autodesk, Inc.

    Abstract: A method and system provide the ability to utilize three-dimensional (3D) models to perform a predictive task. Multiple 3D models, consisting of non-Euclidean data, are obtained. Each 3D model is translated into a relational graph with nodes and edges. Each relational graph is processed using a graph neural network (GNN) that computes a node representation per node. The node representations are aggregated into a structural representation of the 3D model. Multiple different views of the 3D model are captured and passed through a convolutional neural network (CNN) to compute a view representation of each view. The view representations are aggregated into a single visual representation. The GNN and CNN are trained using a multiview contrastive training objective to maximize agreement between the structural representation and the single visual representation to form final learned representations. The final learned representation is utilized to perform the predictive task.

    Three-dimensional generative design based on two-dimensional sketching

    公开(公告)号:US11113865B2

    公开(公告)日:2021-09-07

    申请号:US16520269

    申请日:2019-07-23

    Applicant: AUTODESK, INC.

    Abstract: One embodiment of the present invention provides a technique for generating a three-dimensional model from a two-dimensional sketch. The technique includes receiving input indicating a set of points defining a first sketch element and a second set of points defining a second sketch element included in a sketch. The technique further includes identifying one or more design relationships between the first sketch element and the second sketch element. The technique further includes generating a computer model of the sketch that represents a structure linking the first sketch element and the second sketch element according to the one or more design relationships. The technique further includes outputting the first sketch element, the second sketch element, and the structure for display.

    SYSTEM-LEVEL APPROACH TO GOAL-DRIVEN DESIGN
    3.
    发明申请
    SYSTEM-LEVEL APPROACH TO GOAL-DRIVEN DESIGN 审中-公开
    系统级方法的目标设计

    公开(公告)号:US20170024493A1

    公开(公告)日:2017-01-26

    申请号:US15215520

    申请日:2016-07-20

    Applicant: Autodesk, Inc.

    Abstract: A design application is configured to perform a system-level optimization of a collection of system components. The design application iteratively executes a multi-objective solver to optimize structural and functional relationships between the system components in order to meet global design criteria and generate a system design. The design application initializes the design process by extracting from a knowledge base system templates having taxonomic, structural, or functional attributes relevant to the system design. The design application generates the knowledge base by mining taxonomic, structural, and functional relationships from a corpus of engineering texts.

    Abstract translation: 设计应用程序被配置为执行系统组件集合的系统级优化。 设计应用程序迭代地执行多目标求解器以优化系统组件之间的结构和功能关系,以满足全局设计标准并生成系统设计。 设计应用程序通过从具有与系统设计相关的分类,结构或功能属性的知识库系统模板中提取初始化设计过程。 设计应用通过从工程文本语料库挖掘分类学,结构和功能关系来生成知识库。

    Efficient sensitivity analysis for generative parametric design of dynamic mechanical assemblies

    公开(公告)号:US11620418B2

    公开(公告)日:2023-04-04

    申请号:US15924138

    申请日:2018-03-16

    Applicant: AUTODESK, INC.

    Abstract: A design engine generates a configuration option that includes a specific arrangement of interconnected mechanical elements adhering to one or more design constraints. Each element within a given configuration option is defined by a set of design variables. The design engine implements a parametric optimizer to optimize the set of design variables associated with each configuration option. For a given configuration option, the parametric optimizer discretizes continuous equations governing the physical dynamics of the configuration. The parametric optimizer then determines the gradient of an objective function based on the discretized equations the gradient of objective and constraint functions based on discrete direct differentiation method or discrete adjoint variable method derived directly from the discretized motion equations. Then, the parametric optimizer traverses a design space where the configuration option resides to reduce improve the objective function, thereby optimizing the design variables.

    Automatic design of mechanical assemblies using estimation of distribution algorithm

    公开(公告)号:US11487917B2

    公开(公告)日:2022-11-01

    申请号:US16434085

    申请日:2019-06-06

    Applicant: AUTODESK, INC.

    Abstract: A design engine implements a probabilistic approach to generating designs for computer-aided design (CAD) assemblies. The design engine initially generates a population of designs based on a problem definition associated with a design problem. Each design includes a randomly-generated set of design values assigned to various design variables. The design engine repairs any infeasible designs in the population of designs and then executes a dynamic simulation with the population of designs. The design engine selects the most performant designs and identifies, based on those performant designs, design variables that are dependent on one another. The design engine generates a probability model indicating conditional probabilities between design values associated with dependent design variables. The design engine then iteratively samples the probability model to generate a subsequent population of designs. In this manner, the design engine can automatically generate designs for mechanical assemblies significantly faster than possible with conventional algorithmic design techniques.

    System-level approach to goal-driven design

    公开(公告)号:US10803207B2

    公开(公告)日:2020-10-13

    申请号:US15215520

    申请日:2016-07-20

    Applicant: Autodesk, Inc.

    Abstract: A design application is configured to perform a system-level optimization of a collection of system components. The design application iteratively executes a multi-objective solver to optimize structural and functional relationships between the system components in order to meet global design criteria and generate a system design. The design application initializes the design process by extracting from a knowledge base system templates having taxonomic, structural, or functional attributes relevant to the system design. The design application generates the knowledge base by mining taxonomic, structural, and functional relationships from a corpus of engineering texts.

    Knowledge-based alternative design recommendations

    公开(公告)号:US11301595B2

    公开(公告)日:2022-04-12

    申请号:US15719445

    申请日:2017-09-28

    Applicant: AUTODESK, INC.

    Abstract: Embodiments of the invention disclosed herein provide techniques for generating an alternative design recommendation. The techniques include determining, via a processor, a first function associated with one or more first components included in a first design. The techniques further include analyzing, via the processor, a plurality of components included in a database to select one or more second components associated with a second function that corresponds to the first function. The techniques further include causing an alternative design recommendation that includes the one or more second components to be output for display.

    Knowledge-based design autocomplete recommendations

    公开(公告)号:US11244083B2

    公开(公告)日:2022-02-08

    申请号:US15719449

    申请日:2017-09-28

    Applicant: AUTODESK, INC.

    Abstract: Embodiments of the invention disclosed herein provide techniques for generating a design autocomplete recommendation. The techniques include determining one or more first components included in a first design. The techniques further include analyzing, via the processor, a plurality of designs included in a database to select a second design that includes one or more second components that correspond to the one or more first components. The techniques further include selecting, via a processor, one or more third components included in the second design that are not included in the first design. The techniques further include causing an autocomplete recommendation that includes the one or more third components to be output.

    Techniques for using controlled natural language to capture design intent for computer-aided design

    公开(公告)号:US11113430B2

    公开(公告)日:2021-09-07

    申请号:US14707897

    申请日:2015-05-08

    Applicant: AUTODESK, Inc

    Abstract: A computing device for receiving a design problem statement describing a design problem in a controlled natural language (CNL) that defines permitted lexicons and syntax structures. The design problem statement is processed using the CNL lexicons and syntax structures to produce a job description executable by a design application for generating a design solution for the design problem statement. An improved CNL user interface that assists users to produce valid design problem statements that are CNL-compliant. The CNL user interface receives user-selectable terms that are compliant with the CNL lexicons and generates candidate problem statements that are compliant with CNL syntax structures and receives a selection of a candidate problem statement that is added to the design problem statement. A graphical user interface may display a graphical representation of a design problem statement that can be directly modified. A dialogue-based design process to explore possible design intentions and design solutions.

Patent Agency Ranking