DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 12/02/2024 and 02/24/2026 was filed and is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Amendment
The preliminary amendment filed on 10/25/2024 has been entered. Claims 18-20 were cancelled. Claims 1-17 remain pending in the application.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-9, 11-17 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Wu (“DeepCAD: A Deep Generative Network for Computer-Aided Design Models”, 2021, as cited in IDS filed 02/24/2026).
Regarding claims 1 and 17, Wu teaches A method for configuring a machine learning model (Wu, see nearest image below, “deep generative network” is being interpreted to involve a machine learning model) to generate geometric models (Wu, pg 1, column 2, ¶1, reproduced below:
PNG
media_image1.png
450
1130
media_image1.png
Greyscale
. “construct a 3D shape” is being interpreted as involving a method “to generate geometric models”), the method comprising:
Obtaining (Wu, see nearest image below, “We therefore create a new dataset”, which shows the training obtains data from the dataset) geometric modelling data (Wu, see nearest image below, “We…create a new dataset that…provides CAD command sequences”, which is being interpreted to involve geometric modelling data) comprising sequences of geometric modelling operations (Wu, pg 5, Section 3.3, ¶2-4 and Section 3.4, ¶1, reproduced below:
PNG
media_image2.png
772
560
media_image2.png
Greyscale
. “CAD command sequences” and “commands listed in Table 1” are being interpreted to involve “sequences of geometric modelling operations”); and
training the machine learning model (Wu, see nearest image above, “Leverage the dataset, we train our autoencoder network”) on the geometric modelling data (Wu, see nearest image above, “Leverage the data” is being interpreted to involve “geometric modelling data”) to generate geometric models (Wu, see pg 1, column 2, ¶1 image above, “outputs a sequence of operations used in CAD tools [such as SolidWorks and AutoCAD] to construct a 3D shape”. “Outputs” is being interpreted as involving generating geometric models) encoded as tokenized representations of sequences of geometric modelling operations (Wu, see pg 1, column 2, ¶1 image above, “outputs a sequence of operations used in CAD tools”. Which is being interpreted to involve tokenized representations of sequences of geometric modelling operations) to be performed to build the geometric models (Wu, see pg 1, column 2, ¶1 image above, “outputs a sequence of operations used in CAD tools [such as SolidWorks and AutoCAD] to construct a 3D shape”. “Construct a 3D shape” is being interpreted as involving building the geometric models);
wherein the machine learning model (Wu, pg 5, Section 3.4, ¶1, “Leveraging the dataset, we train our autoencoder”. “Autoencoder” is being interpreted to involve the machine learning model) is trained to generate the geometric models (Wu, see pg 1, column 2, ¶1 image above, “outputs a sequence of operations used in CAD tools [such as SolidWorks and AutoCAD] to construct a 3D shape”. “Outputs” is being interpreted as involving generating geometric models) in accordance with learned geometric modelling practices (Wu, pg 5, Section 3.3 to 3.4 image above, Section 3.4, “Leveraging the dataset, we train our autoencoder”. “Train” is being interpreted to involving “learning”) extracted from the geometric modelling data (Wu, pg 5, Section 3.3 to 3.4 image above, “In the end, we collect a dataset with 178, 238 CAD designs all described as CAD command sequences”).
Regarding claim 2, Wu teaches The method of claim 1, wherein the sequences of geometric modelling operations of the geometric modelling data (Wu, pg 1, column 2, ¶1, reproduced below:
PNG
media_image1.png
450
1130
media_image1.png
Greyscale
. “sequence of operations used in CAD tools” is being interpreted to involve “sequences of geometric modelling operations of the geometric modelling data”) comprises training data derived from user input into a geometric modelling tool (Wu, supplementary document, Section A, ¶1, reproduced here:
PNG
media_image3.png
222
562
media_image3.png
Greyscale
. “User-created mechanical parts” are being interpreted as involving user input into a geometric modelling tool. “Dataset” is being interpreted as being used for training).
Regarding claim 3, Wu teaches The method of claim 2, wherein a tokenized representation of a geometric model generated by the machine learning model comprises:
a plurality of coordinate tokens representing vertices of the geometric model (Wu, see nearest image below, the Line row has the x, y parameters, which are being interpreted as involving example coordinate tokens that represent vertices of the geometric model); and
one or more operation tokens (Wu, see nearest image below, the CAD commands are being interpreted as involving “one or more operation tokens”) representing one or more geometric modelling operations (Wu, see nearest image below, row 1, the line command is being interpreted as an example of one or more geometric modelling operations) involving one or more of the vertices of the geometric model (Wu, pg 3, Table 1, reproduced below:
PNG
media_image4.png
670
744
media_image4.png
Greyscale
. “Line end-point” from the second row is being interpreted as an example “one or more vertices of the geometric model”).
Regarding claim 4, Wu teaches The method of claim 3, wherein the geometric model comprises a plurality of geometric entities (Wu, Figure 2, reproduced below:
PNG
media_image5.png
760
754
media_image5.png
Greyscale
. The Extrude 1 and Extrude 2 stages are being interpreted as example plurality of geometric entities), and
wherein the tokenized representation encodes for a geometric modelling operation (Wu, see nearest image above, the command sequence shown is being interpreted as involving the tokenized representation that encodes for a geometric modelling operation) that involves selecting a geometric entity encoded for earlier in the tokenized representation (Wu, see nearest image above, Sketch 1 is being interpreted as an example geometric entity encoded for earlier as Extrude 1 is involved with Sketch 1. Sketch 1 extrusion to Extrude 1 is being interpreted as involving being selected so that the extrusion happens on the selected entity.).
Regarding claim 5, Wu teaches The method of claim 4, wherein the selection of the geometric entity involves defining one or more coordinates that correspond to the geometric entity (Wu, See Figure 2 image above, Parameterized command sequence, “L_2: (2, 0)” is being interpreted as an example geometric entity that involves defining one or more coordinates that correspond to the geometric entity).
Regarding claim 6, Wu teaches The method of claim 4, wherein the geometric modelling operation involves transforming the geometric entity (Wu, see Figure 2 image above, the extrusion from Sketch 1 to Extrude 1 is being interpreted as involving transforming the geometric entity Sketch 1 into Extrude 1).
Regarding claim 7, Wu teaches The method of claim 4, wherein the geometric modelling operation involves defining an attribute of a first geometric entity with respect to a second geometric entity (Wu, see Figure 2 image above, “L_4: (0,2)” is being interpreted as an example attribute for a first geometric entity with respect to a second geometric entity, “L_5: (0, 0)”.).
Regarding claim 8, Wu teaches The method of claim 7, wherein the attribute is a geometric constraint (Wu, see Figure 2 image above, the Sketch 1 to Extrude 1 extrusion height is being interpreted as an attribute for a geometric constraint. Otherwise, the extrusion would go into infinity or not work at all).
Regarding claim 9, Wu teaches The method of claim 1, wherein the learned geometric modelling practices extracted from the geometric modelling data comprise tendencies to apply different geometric modelling techniques (Wu, see Figure 2 image above, line creation and circle creation are being interpreted as example geometric modelling techniques) in different geometric modelling scenarios (Wu, see Figure 2, Sketch 1 and Sketch 2 are being interpreted as involving different scenarios).
Regarding claim 11, Wu teaches The method of claim 1, wherein the machine learning model is configured to apply self-attention (Wu, Abstract: “We propose a CAD generative network based on the Transformer”; Wu, pg 2, column 2, Subsection “Transformer-based models”, ¶1: “what was introduced as an attention-based building block for many natural language processing tasks [13]”. Reference 13, as one with ordinary skill would know, has self-attention) among the elements of the tokenized representation (Wu, Section 3.2. Autoencoder for CAD models, ¶1: “We now introduce an autoencoder network that leverages our representation of CAD commands.” Which is being interpreted to involve the elements of the tokenized representation).
Regarding claim 12, Wu teaches The method of claim 11, wherein the machine learning model is further configured to apply cross-attention (Wu, Abstract: “We propose a CAD generative network based on the Transformer”; Wu, pg 2, column 2, Subsection “Transformer-based models”, ¶1: “what was introduced as an attention-based building block for many natural language processing tasks [13]”. Reference 13, as one with ordinary skill would know, has cross-attention. As seen in Section 3.2, ¶1 from reference 13) between the elements of the tokenized representation and a context token (Wu, Section 3.2, ¶1-2, reproduced below:
PNG
media_image6.png
438
562
media_image6.png
Greyscale
. “Embedding” is being interpreted to involve tokenized representation. CAD example command from Table 1, Circle, is being interpreted as involving “context token”).
Regarding claim 13, Wu teaches A method for generating geometric models (Wu, see nearest image below, “deep generative network that outputs a sequence of operations used in CAD tools”), the method comprising:
applying a machine learning model (Wu, see nearest image below, “Output” from the “deep generative network” is being interpreted as involving applying a machine learning model) to generate a tokenized representation of a geometric model (Wu, Section 3.2, ¶2, “each command…is projected onto a continuous embedding space” is being interpreted to involve “tokenized representation”. The command is involved with CAD command, which involves “a geometric model”), wherein the tokenized representation of the geometric model (Wu, see nearest image below, “deep generative network” is being interpreted to involve “tokenized representation of the geometric model”) defines a sequence of geometric modelling operations (Wu, see nearest image below, “operational sequence” is being interpreted as involving “a sequence of geometric modelling operations”) that is to be performed to build the geometric model (Wu, pg 1, column 2, ¶1, reproduced below:
PNG
media_image1.png
450
1130
media_image1.png
Greyscale
. “construct a 3D shape” is being interpreted as involving a method “to build the geometric model”. “Output” from the “deep generative network” is being interpreted as involving applying a machine learning model.).
Regarding claim 14, Wu teaches The method of claim 13, further comprising:
converting the tokenized representation (Wu, see nearest image below, “embedding” which is related to latent vector z, is being interpreted to involve “tokenized representation”. The decoder that takes latent vector z and outputs a generated CAD command sequence is being interpreted as involving “converting”) of the geometric model into a format suitable for use in a geometric modelling environment (Wu, Section 3.2, ¶1-2, reproduced below:
PNG
media_image7.png
469
560
media_image7.png
Greyscale
. “CAD command sequences” is being interpreted as involving “format suitable for use in a geometric modelling environment”, as CAD is being interpreted to involve “geometric modelling environment”); and
instantiating the geometric model in a geometric modelling environment (Wu, pg 1, column 2, ¶1, “We propose a deep generative network that outputs a sequence of operations used in CAD tools (such as SolidWorks and AutoCAD) to construct a 3D shape”. Which shows it would be obvious to instantiate the geometric model in a geometric modelling environment).
Regarding claim 15, Wu teaches The method of claim 13, wherein the machine learning model is trained (Wu, Section 3.2, ¶1, the autoencoder network is being interpreted as “the machine learning model is trained”) to generate tokenized representations (Wu, Section 3.2, ¶2, “embedding” is being interpreted as involving “tokenized representations”, the encoding processing is being interpreted as involving generating) of geometric models of particular object classes (Wu, Table 1, line, arc, circle, commands are being interpreted as examples of “geometric models of particular object classes”).
Regarding claim 16, Wu teaches The method of claim 13, wherein the machine learning model is trained to generate tokenized representations of building structures (Wu, Table 1, Line, Arc, and Circle commands are being interpreted as involving “generate tokenized representations of building structures”).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wu, in view of Nash (“PolyGen: An Autoregressive Generative Model of 3D Meshes”, 2020, as cited in IDS filed 02/24/2026).
Regarding claim 10, Wu teaches The method of claim 1, wherein the machine learning model
However, Wu does not appear to explicitly teach an autoregressive generative model.
Pertaining to the same field of endeavor, Nash teaches
comprises an autoregressive generative model (Nash, pg 1, column 2, first partial paragraph: “….we present PolyGen, a neural generative model of meshes that autoregressively estimates a joint distribution over mesh vertices and faces”).
Wu and Nash are considered to be analogous art because they are directed to machine learning model generation of geometry. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the method and system for machine learning model generation of geometry (as taught by Wu) to include, or try, an autoregressive generative model (as taught by Nash) because the combination provides an improvement to the quality of models (Nash, Abstract).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Wang et al (US 10769848 B1) discloses machine learning model to generate geometric models (interpreted from mesh representation).
Zhang et al (“Point-M2AE: Multi-scale Masked Autoencoders for Hierarchical Point Cloud Pre-training”, May 2022) discloses machine learning model that generates tokenized representation of geometry (see Figure 1).
Fan et al (“CADTransformer: Panoptic Symbol Spotting Transformer for CAD Drawings”, June 2022) discloses machine learning model that generates tokenized representation of CAD geometry.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOHNNY B DUONG whose telephone number is (571)272-1358. The examiner can normally be reached Monday - Thursday 10a-9p (ET).
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Matthew Bella can be reached at (571)272-7778. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/J.B.D./Examiner, Art Unit 2667
/MATTHEW C BELLA/Supervisory Patent Examiner, Art Unit 2667
Prosecution Insights