Virtual generation of labeled motion sensor data

Patent metadata

Field	Value
Publication number	US-10657656-B2
Application number	US-201816010190-A
Country	US
Kind code	B2
Filing date	Jun 15, 2018
Priority date	Jun 15, 2018
Publication date	May 19, 2020
Grant date	May 19, 2020

Abstract

Systems, computer-implemented methods, and computer program products to generate virtual motion sensor data from computer animations are provided. A system can comprise a memory that stores computer executable components and a processor that executes the computer executable components stored in the memory. The computer executable components can comprise a tracker component that can track virtual location data corresponding to a feature of a computer animated character in a virtual environment. The computer executable components can further comprise a virtual motion sensor component that, based on the virtual location data, can generate virtual motion sensor data.

First claim

What is claimed is:

1. A system, comprising:
a memory that stores computer executable components; and
a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a tracker component that tracks virtual location data corresponding to a feature of a computer animated character in a virtual environment;
a virtual motion sensor component that, based on the virtual location data, generates virtual motion sensor data; and
a model trainer component that, based on the virtual motion sensor data, employs machine learning to train a predictive model to identify one or more movement activities of an entity, wherein the predictive model is trained to dynamically control an amount of random variation to apply to the virtual motion sensor data to identify the one or more movement activities of the entity within a defined range of acceptable variation from the virtual motion sensor data.

2. The system of claim 1 , wherein the virtual motion sensor data is selected from the group consisting of linear velocity, linear acceleration, angular velocity, and angular acceleration.

3. The system of claim 1 , further comprising a feature definition component that defines at least one of: a position or an orientation of the feature relative to the computer animated character.

4. The system of claim 1 , further comprising an animation collection component that collects desired animation movements from pre-existing humanoid computer animations, thereby facilitating at least one of improved performance, improved efficiency, or improved available storage capacity associated with the memory.

5. The system of claim 1 , wherein the virtual location data is selected from at least one of: position data or rotation data.

6. The system of claim 1 , wherein a virtual sensor is associated with a location of interest of an avatar.

7. The system of claim 1 , wherein the computer animated character corresponds to one or more ground truth labels, thereby facilitating at least one of improved processing capacity, improved processing performance, improved processing efficiency, or improved processing time associated with the processor.

8. The system of claim 1 , wherein a set of collected computer animations are divided into a first subset of animations that contain desired motor events that can simulate a desired animation movement and a second subset of animations that contain undesired motor events that cannot simulate the desired animation movement, wherein the first subset of animations are positively labeled and the second subset of animations negatively labeled.

9. A computer-implemented method, comprising:
tracking, by a system operatively coupled to a processor, virtual location data corresponding to a feature of a computer animated character in a virtual environment;
based on the virtual location data, generating, by the system, virtual motion sensor data;
based on the virtual motion sensor data, employing, by the system, machine learning to train a predictive model to identify one or more movement activities of an entity, wherein the predictive model is trained to dynamically control an amount of random variation to apply to the virtual motion sensor data to identify the one or more movement activities of the entity within a defined range of acceptable variation from the virtual motion sensor data.

10. The computer-implemented method of claim 9 , wherein the virtual motion sensor data is selected from the group consisting of linear velocity, linear acceleration, angular velocity, and angular acceleration.

11. The computer-implemented method of claim 9 , further comprising defining, by the system, at least one of a position or orientation of the feature relative to the computer animated character.

12. The computer-implemented method of claim 9 , further comprising collecting, by the system, desired animation movements from pre-existing humanoid computer animations, thereby facilitating at least one of improved performance, improved efficiency, or improved available storage capacity associated with the memory.

13. The computer-implemented method of claim 9 , wherein the virtual location data is selected from at least one of: position data or rotation data.

14. The computer-implemented method of claim 9 , wherein the computer animated character corresponds to one or more ground truth labels, thereby facilitating at least one of improved processing capacity, improved processing performance, improved processing efficiency, or improved processing time associated with the processor.

15. The computer-implemented method of claim 9 , wherein a virtual sensor is associated with location of interest of an avatar.

16. A computer program product facilitating a virtual generation of motion sensor data, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
track, by the processor, virtual location data corresponding to a feature of a computer animated character in a virtual environment;
based on the virtual location data, generate, by the processor, virtual motion sensor data; and
based on the virtual motion sensor data, employ, by the processor, machine learning to train a predictive model to identify one or more movement activities of an entity, wherein the predictive model is trained to dynamically control an amount of random variation to apply to the virtual motion sensor data to identify the one or more movement activities of the entity within a defined range of acceptable variation from the virtual motion sensor data.

17. The computer program product of claim 16 , wherein the virtual motion sensor data is selected from the group consisting of linear velocity, linear acceleration, angular velocity, and angular acceleration.

18. The computer program product of claim 16 , wherein the program instructions are further executable by the processor to cause the processor to:
define, by the processor, at least one of position or orientation of the feature relative to the computer animated character.

19. The computer program product of claim 16 , wherein the program instructions are further executable by the processor to cause the processor to:
collect, by the processor, desired animation movements from pre-existing humanoid computer animations, thereby facilitating at least one of improved performance, improved efficiency, or improved available storage capacity associated with the memory.

20. The computer program product of claim 16 , wherein the computer animated character corresponds to one or more ground truth labels, thereby facilitating at least one of improved processing capacity, improved processing performance, improved processing efficiency, or improved processing time associated with the processor.

Assignees

• IBM

Inventors

• Cavallo Marco
• Tejwani Ravi
• Watson Patrick
• Sipolins Aldis
• Reinen Jenna
• Wu Hui

Classifications

• G06F3/011

  Arrangements for interaction with the human body, e.g. for user immersion in virtual reality (blind teaching G09B21/00) · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06F16/78

  Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually · CPC title

• G06T7/246 Primary

  using feature-based methods, e.g. the tracking of corners or segments · CPC title

• G06T13/40

  of characters, e.g. humans, animals or virtual beings · CPC title