Autodesk, Inc.

Fraunhofer Center for Research in Computer Graphics, Inc.

Michael Haley

Thomas Davies

Morgan Fabian

Ara Danielyan

Erin Marie Bradner

Yoshihito Yotto Koga

Shelly Mujtaba

Anup Chatterjee

Kirill Goldin

Tara A Adiseshan

Laura Quiroga

Lars B Karle

Pantelis Katsiaris

Senthil Gandhi

In various embodiments, a training application generates a trained encoder that automatically generates shape embeddings having a first size and representing three-dimensional (3D) geometry shapes. First, the training application generates a different view activation for each of multiple views associated with a first 3D geometry based on a first convolutional neural network (CNN) block. The training application then aggregates the view activations to generate a tiled activation. Subsequently, the training application generates a first shape embedding having the first size based on the tiled activation and a second CNN block. The training application then generates multiple re-constructed views based on the first shape embedding. The training application performs training operation(s) on at least one of the first CNN block and the second CNN block based on the views and the re-constructed views to generate the trained encoder.

Shaped-based techniques for exploring design spaces

A model generator implements a data-driven approach to generating a robot model that describes one or more physical properties of a robot. The model generator generates a set of basis functions that generically describes a range of physical properties of a wide range of systems. The model generator then generates a set of coefficients corresponding to the set of basis functions based on one or more commands issued to the robot, one or more corresponding end effector positions implemented by the robot, and a sparsity constraint. The model generator generates the robot model by combining the set of basis functions with the set of coefficients. In doing so, the model generator disables specific basis functions that do not describe physical properties associated with the robot. The robot model can subsequently be used within a robot controller to generate commands for controlling the robot.

Techniques for generating controllers for robots

In various embodiments, a training application generates a trained encoder that automatically generates shape embeddings having a first size and representing three-dimensional (3D) geometry shapes, First, the training application generates a different view activation for each of multiple views associated with a first 3D geometry based on a first convolutional neural network (CNN) block. The training application then aggregates the view activations to generate a tiled activation. Subsequently, the training application generates a first shape embedding having the first size based on the tiled activation and a second CNN block. The training application then generates multiple re-constructed views based on the first shape embedding. The training application performs training operation(s) on at least one of the first CNN block and the second CNN block based on the views and the re-constructed views to generate the trained encoder.

A method, apparatus, system, and computer program product provide the ability to define an attribute of a design part in a manufacturing or construction project. Apart set of two or more design parts is obtained. Each of the two or more design parts includes metadata. An external corpus is mined to determine a text set of terms relevant to the metadata. The metadata is mined for a pattern based on the text set to discover a metadata property. An attribute is defined based on the metadata property. A graphical user interface is used to search for design parts utilizing a filter based on the metadata property.

Graph

A method, apparatus, and article of manufacture provide the ability to manipulate a graphical representation of a real-world object in a computer drawing application. A semantic behavior is defined for the real-world object. A graphical representation of the real-world object (referred to as a subject object) is obtained. The semantic behavior is assigned to the subject object. The subject object is placed into a drawing. When placed, the subject object automatically, without additional user input, places itself into the drawing based on the semantic behavior.

Smart content and placement in a computer-aided design application

A method, apparatus, and computer program product provide the ability to locate and display data in a taxonomy. A canonical taxonomy is defined (that is not exposed to a user that is searching for parts) by defining canonical categories and canonical attributes in a hierarchical structure. Mapping taxonomies are defined that define mapping categories that map to both canonical categories and canonical categories plus canonical attributes. Part data is processed based on the mapping taxonomies during which the data is mapped to a canonical category. A search request is specified in a particular mapping taxonomy format and results are returned (and displayed) that include data from multiple different taxonomies.

Taxonomy based indexing and searching

A technique for managing conference state. Endpoints of the conference are application processes ( ) running on computer systems that are connected by a WAN ( ). Each of the application processes ( ) maintains endpoint state for the conference. A session manager process ( ) in each of the computer systems maintains session manager state for each of the conferences that has an endpoint on the computer system. The session manager conference state includes a copy of the endpoint state for each of the conferences and the session manager ( ) uses a locking mechanism to that the copies of the session manager conference state in all of the session managers ( ) are identical. When an endpoint changes its endpoint state, it informs the session manager ( ), the session manager ( ) incorporates the change into its session manager conference state, and when the locking mechanism permits, exports the change to the session managers ( ) for all the other endpoints. When the session manager ( ) receives a change, it incorporates the change into its session manager conference state.

Growing community of inventors

San Rafael, CA, United States of America

Michael Haley