IP Library Granted Patent US 10,997,796
Granted Patent B2
US 10,997,796 · App. 16/471,965 · Granted May 4, 2021

Identification and redesign of critical thin segments below 3D printer resolution

Inventors: Prakhar Jaiswal (Buffalo, NY); Suraj Ravi Musuvathy (Princeton, NJ); Erhan Arisoy (Princeton, NJ); David Madeley (Louth, GB)
Assignee: Siemens Industry Software Inc.
G06T19/20B29C64/393B33Y50/02B33Y80/00G05B13/0265G05B19/4099G06T5/002G06T5/30G05B2219/35134G05B2219/49008G06T2207/20081G06T2219/2016G06T2219/2021
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 10,997,796
App. No.
16/471,965
Granted
May 4, 2021
Kind
B2
Abstract

Systems and methods may support identification and redesign of critical thin segments in a 3D model that are below 3D printer resolution. Identification of critical thin segments may include segmenting cross-sectional slices of the 3D model into printable segments and non-printable segments and using a machine learning model trained using geometrical features computed on thin regions to classify the non-printable segments as critical or non-critical. Redesign of critical thin segments may include thickening the critical thin segments such that the segment size of the critical thin segments satisfy a thickening criterion with respect to the printer resolution and smoothing sharp corners added to the cross-sectional slice at an intersection between the critical thin segment and a neighboring printable segment. Redesign of the critical thin segments may account for tolerable overhang.

Claims (60)

1. A method comprising:

accessing a three-dimensional (3D) model representative of an object to be constructed by a 3D printer;

extracting a cross-sectional slice of the 3D model perpendicular to a build direction for construction of the object by the 3D printer;

segmenting the cross-sectional slice into printable segments and non-printable segments, wherein the non-printable segments have a segment size smaller than a printer resolution of the 3D printer;

using a machine learning model trained using geometrical features computed on thin regions to classify each of the non-printable segments as a critical non-printable segment or a non-critical non-printable segment;

modifying the 3D model by thickening the critical non-printable segments such that the segment size of the critical non-printable segments satisfy a thickening criterion with respect to the printer resolution;

determining not to redesign the non-critical non-printable segments in the 3D model even though the non-critical non-printable segments have a segment size smaller than a printer resolution of the 3D printer; and

providing the modified 3D model to the 3D printer for construction of the object,

wherein the determining not to redesign the non-critical non-printable segments in the 3D model comprises including the non-critical non-printable segments in the modified 3D model.

2. The method of claim 1 , wherein thickening the critical non-printable segments comprises, for a given critical non-printable segment:

incrementally thickening the given critical non-printable segment with a dilation shape over a number of dilation iterations; and

incrementally increasing the thickening amount over successive dilation iterations based on a function of the number of iterations and the printer resolution.

3. The method of claim 1 , wherein modifying the 3D model further comprises, for a given critical non-printable segment:

identifying, at an intersection between the critical non-printable segment and a neighboring printable segment, a sharp corner added to the cross-sectional slice by the thickening; and

smoothing the sharp corner through curve interpolation.

4. The method of claim 1 , wherein modifying the 3D model by thickening the critical non-printable segments comprises, for a given critical non-printable segment:

modifying the 3D model by thickening the given critical thin-non-printable segment with an inverted conical thickening shape with a conical slant angle determined based on an allowable overhang parameter specified for construction of the object with the 3D printer.

5. The method of claim 4 , wherein the inverted conical thickening shape is 3D, has a reference point at a center of a cone base of the inverted conical thickening shape and the cone base has a diameter equal to a diameter associated with a 2D representation of the printer resolution.

6. The method of claim 4 , wherein modifying the 3D model by thickening the critical thin segments comprises, for the given critical non-printable segment, modifying a voxel representation of the 3D model by dilating the given critical non-printable segment with the inverted conical thickening shape; and

further comprising converting the modified voxel representation of the 3D model into a triangulated or CAD surface representation of the object.

7. A non-transitory machine-readable medium comprising instructions executable by a processor to:

access a three-dimensional (3D) model representative of an object to be constructed by a 3D printer;

extract a cross-sectional slice of the 3D model perpendicular to a build direction for construction of the object by the 3D printer;

segment the cross-sectional slice into printable segments and non-printable segments, wherein the non-printable segments have a segment size smaller than a printer resolution of the 3D printer;

classify each of the non-printable segments as a critical non-printable segment or a non-critical non-printable segment;

modify the 3D model by thickening the critical non-printable segments such that the segment size of the critical non-printable segments satisfy a thickening criterion with respect to the printer resolution;

determine not to redesign the non-critical non-printable segments in the 3D model even though the non-critical non-printable segments have a segment size smaller than a printer resolution of the 3D printer, wherein determination not to redesign the non-critical non-printable segments in the 3D model comprises modifying the 3D model to remove the non-critical non-printable segments; and

provide the modified 3D model to the 3D printer for construction of the object.

8. The non-transitory machine-readable medium of claim 7 , wherein the instructions executable to thicken the critical non-printable segments comprise instructions executable to, for a given critical non-printable segment:

incrementally thicken the given critical non-printable segment with a dilation shape over a number of dilation iterations; and

incrementally increase a size of the dilation shape over successive dilation iterations based on a function of the number of iterations and the printer resolution.

9. The non-transitory machine-readable medium of claim 7 , wherein the instructions further comprise instructions executable to smooth a sharp corner at an intersection between a given critical non-printable segment and a neighboring printable segment, including by:

identifying a region surrounding the sharp corner with a region size determined based on a factor of the printer resolution; and

applying a curve interpolation in the identified region surrounding the sharp corner to smooth the sharp corner.

10. The non-transitory machine-readable medium of claim 7 , wherein the thickening criterion is satisfied when the segment size of the thickened critical thin segments exceed a threshold percentage of the printer resolution.

11. The non-transitory machine-readable medium of claim 7 , wherein the instructions to modify the 3D model comprise instructions further to:

extend the cross-sectional slice by a slice height of the cross-sectional slice; and

merge the extended cross-sectional slice with other extended cross-sectional slices to form the modified 3D model.

12. A system comprising:

a thin segment identification engine configured to:

access a three-dimensional (3D) model representative of an object to be constructed by a 3D printer;

extract cross-sectional slices of the 3D model perpendicular to a build direction for construction of the object by the 3D printer;

segment the cross-sectional slices into printable segments and non-printable segments;

use machine learning model trained using geometrical features computed on thin regions to classify each of the non-printable segments as a critical non-printable segment or a non-critical non-printable segment; and

a thin segment redesign engine configured to:

modify the 3D model by thickening the critical non-printable segments such that the segment size of the critical non-printable segments satisfy a thickening criterion with respect to the printer resolution;

determine not to redesign the non-critical non-printable segments in the 3D model even though the non-critical non-printable segments have a segment size smaller than a printer resolution of the 3D printer, wherein the thin segment redesign engine is configured to determine not to redesign the non-critical non-printable segments in the 3D model by modifying the 3D model to remove the non-critical non-printable segments; and

provide the modified 3D model to the 3D printer for construction of the object.

13. The system of claim 12 , wherein the thin segment redesign engine is further configured to, after thickening the critical non-printable segments:

extend the cross-sectional slices by a slice height of the cross-sectional slices along the build direction, opposite the build direction, or along a combination of both; and

merge the extended cross-sectional slices to form the modified 3D model.

14. The system of claim 12 , wherein the thin segment redesign engine is configured to thicken the critical non-printable segments by, for a given critical non-printable segment:

incrementally thickening the given critical non-printable segment with a dilation shape over a number of dilation iterations; and

incrementally increasing a size of the dilation shape over successive dilation iterations based on a function of the number of iterations and the printer resolution.

15. The system of claim 12 , wherein the thin segment redesign engine is further configured to smooth a sharp corner at an intersection between a given critical non-printable segment and a neighboring printable segment, including by:

identifying a region surrounding the sharp corner with a region size determined based on a factor of the printer resolution; and

applying a curve interpolation in the identified region surrounding the sharp corner to smooth the sharp corner.

16. The system of claim 12 , wherein the thickening criterion is satisfied when the segment size of the thickened critical thin segments exceed a threshold percentage of the printer resolution.

17. The system of claim 12 , wherein the thin segment redesign engine is configured to thicken the critical non-printable segments with an inverted conical thickening shape with a conical slant angle determined based on an allowable overhang parameter specified for construction of the object with the 3D printer.

18. The system of claim 17 , wherein the inverted conical thickening shape is 3D, has a reference point at a center of a cone base of the inverted conical thickening shape and the cone base has a diameter equal to a diameter associated with a 2D representation of the printer resolution.

Assignments (6)
CHANGE OF NAME Recorded Dec 3, 2019
From: SIEMENS PRODUCT LIFECYCLE MANAGEMENT SOFTWARE INC.
To: SIEMENS INDUSTRY SOFTWARE INC.
Reel/Frame 051171/0024 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 30, 2019
From: JAISWAL, PRAKHAR; MUSUVATHY, SURAJ RAVI; ARISOY, ERHAN
To: SIEMENS CORPORATION
Reel/Frame 050563/0895 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 30, 2019
From: SIEMENS CORPORATION
To: SIEMENS AKTIENGESELLSCHAFT
Reel/Frame 050564/0465 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 30, 2019
From: MADELEY, DAVID
To: SIEMENS INDUSTRY SOFTWARE LIMITED
Reel/Frame 050564/0722 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 30, 2019
From: SIEMENS INDUSTRY SOFTWARE LIMITED
To: SIEMENS PRODUCT LIFECYCLE MANAGEMENT SOFTWARE INC.
Reel/Frame 050565/0767 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 30, 2019
From: SIEMENS AKTIENGESELLSCHAFT
To: SIEMENS PRODUCT LIFECYCLE MANAGEMENT SOFTWARE INC.
Reel/Frame 050565/0984 →