Systems and methods for predicting responses of a particle to a stimulus

Patent metadata

Field	Value
Publication number	US-11423196-B2
Application number	US-201816202719-A
Country	US
Kind code	B2
Filing date	Nov 28, 2018
Priority date	Nov 28, 2018
Publication date	Aug 23, 2022
Grant date	Aug 23, 2022

Abstract

System, methods, and other embodiments described herein relate to predicting effects of a stimulus on a particle or other material structure. In one embodiment, a method includes receiving a segmented image of a particle that identifies at least semantics of the particle and associated characteristics according to subregions of the particle. The method includes analyzing, using a stimulus model, the segmented image to predict changes in the particle associated with applying the stimulus to the particle. Analyzing the segmented image includes generating a predicted image identifying characteristics, semantics and other properties of the particle according to the changes. The method includes providing the predicted image as an electronic output.

First claim

What is claimed is:

1. A semantics system for predicting effects of a stimulus on a particle, comprising:
one or more processors;
a memory communicably coupled to the one or more processors and storing:
a segmentation module including instructions that when executed by the one or more processors cause the one or more processors to receive a segmented image of the particle that identifies at least semantics of the particle and associated characteristics according to subregions of the particle; and
a prediction module including instructions that when executed by the one or more processors cause the one or more processors to analyze, using a stimulus model, the segmented image to predict changes in the particle associated with applying the stimulus to the particle,
wherein the prediction module includes instructions to generate a predicted image of the particle according to the changes, and
wherein the prediction module includes instructions to provide the predicted image as an electronic output.

2. The semantics system of claim 1 , wherein the stimulus is a stress that effects a physical structure of the particle including the semantics and the associated characteristics,
wherein the prediction module includes instructions to analyze the segmented image using the stimulus model to predict the changes without applying the stress to the particle, and
wherein the prediction module includes instructions to generate the predicted image including instructions to generate the predicted image with segments identifying modified subregions according to the stimulus.

3. The semantics system of claim 1 , wherein the stimulus model is a machine learning algorithm that accepts the segmented image as an electronic input and generates the predicted image to simulate effects of the stimulus on the particle.

4. The semantics system of claim 1 , wherein the prediction module includes instructions to analyze the segmented image to predict the changes including instructions to predict modifications to the semantics through inferences embodied in the stimulus model about how at least the associated characteristics of the particle respond to the stimulus, and
wherein the stimulus includes one of heat, mechanical stress, chemical exposure, and electrochemical effects, and wherein the stimulus model is trained to predict the changes for a single stress.

5. The semantics system of claim 1 , wherein the segmentation module includes instructions to receive a particle image from a transmission electron microscopy (TEM) microscope that scans the particle to produce the particle image,
wherein the particle image includes diffraction patterns that are patterns of electrons as scattered onto a detector in the TEM microscope resulting from the electrons interacting with the particle at a location corresponding with a respective pixel of pixels in the particle image.

6. The semantics system of claim 5 , wherein the segmentation module includes instructions to analyze, using a semantics model that performs semantic segmentation, the particle image to produce the segmented image by identifying the associated characteristics of the particle associated with pixels of the particle image, and
wherein the segmentation module includes instructions to identify the semantics of the particle from the segmented image according to relationships between the subregions that define the semantics for the particle.

7. The semantics system of claim 1 , wherein the stimulus model is a generative adversarial network (GAN).

8. The semantics system of claim 1 , wherein the prediction module includes instructions to train the stimulus model using pairs of segmented images for a plurality of particles depicting the particles before and after being subjected to the stimulus.

9. A non-transitory computer-readable medium for predicting effects of a stimulus on a particle and including instructions that when executed by one or more processors cause the one or more processors to:
receive a segmented image of the particle that identifies at least semantics of the particle and associated characteristics according to subregions of the particle;
analyze, using a stimulus model, the segmented image to predict changes in the particle associated with applying the stimulus to the particle, wherein the instructions to predict the changes include instructions to generate a predicted image of the particle according to the changes; and
provide the predicted image as an electronic output.

10. The non-transitory computer-readable medium of claim 9 , wherein the stimulus is a stress that effects a physical structure of the particle including the semantics and the associated characteristics,
wherein the instructions include instructions to analyze the segmented image using the stimulus model to predict the changes without applying the stress to the particle, and
wherein the instructions include instructions to generate the predicted image including instructions to generate the predicted image with segments identifying modified subregions according to the stimulus.

11. The non-transitory computer-readable medium of claim 9 , wherein the stimulus model is a machine learning algorithm that accepts the segmented image as an electronic input and generates the predicted image to simulate effects of the stimulus on the particle.

12. The non-transitory computer-readable medium of claim 9 , wherein the instructions to analyze the segmented image to predict the changes including instructions to predict modifications to the semantics through inferences embodied in the stimulus model about how at least the associated characteristics of the particle respond to the stimulus, and
wherein the stimulus includes one of heat, mechanical stress, chemical exposure, and electrochemical effects, and wherein the stimulus model is trained to predict the changes for a single stress.

13. The non-transitory computer-readable medium of claim 9 , wherein the instructions include instructions to receive a particle image from a transmission electron microscopy (TEM) microscope that scans the particle to produce the particle image,
wherein the particle image includes diffraction patterns that are patterns of electrons as scattered onto a detector in the TEM microscope resulting from the electrons interacting with the particle at a location corresponding with a respective pixel of pixels in the particle image.

14. A method of predicting effects of a stimulus on a particle, comprising:
receiving a segmented image of a particle that identifies at least semantics of the particle and associated characteristics according to subregions of the particle;
analyzing, using a stimulus model, the segmented image to predict changes in the particle associated with applying the stimulus to the particle,
wherein analyzing the segmented image includes generating a predicted image of the particle according to the changes; and
providing the predicted image as an electronic output.

15. The method of claim 14 , wherein the stimulus is a stress that effects a physical structure of the particle including the semantics and the associated characteristics, wherein analyzing the segmented image using the stimulus model predicts the changes without applying the stress to the particle, and
wherein generating the predicted image includes generating the predicted image with segments identifying modified subregions according to the stimulus.

16. The method of claim 14 , wherein the stimulus model is a machine learning algorithm that accepts the segmented image as an electronic input and generates the predicted image t

Assignees

• Toyota Res Inst Inc

Inventors

• Herring Patrick K

Classifications

• G06F30/25 Primary

  using particle-based methods · CPC title

• G06V20/695

  Preprocessing, e.g. image segmentation · CPC title

• G06V10/26

  Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion · CPC title

• G06V20/698

  Matching; Classification · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title