System and method for generating synthetic container monitoring training data
AssigneeBarrel Proof Technologies LLC
InventorBrian Richard Anderson
Life of this patent
Estimating the term…
Abstract
A method for generating synthetic training data for a container monitoring system includes defining a virtual container having specified geometry and material properties within a digital twin simulation environment, configuring one or more virtual sensors within the digital twin simulation environment, executing a physics simulation to model electromagnetic wave propagation through the virtual container and interaction with a virtual enclosed medium, generating raw synthetic sensor data based on the physics simulation, labeling the raw synthetic sensor data with ground truth parameters, and outputting a labeled training dataset for training a machine learning model based on the labeled raw synthetic sensor data.
Claims30
A method for generating synthetic training data for a container monitoring system, comprising:
- defining, within a digital twin simulation environment, a virtual container having specified geometry and material properties;
- configuring one or more virtual sensors within the digital twin simulation environment;
- executing a physics simulation to model electromagnetic wave propagation through the virtual container and interaction with a virtual enclosed medium;
- generating raw synthetic sensor data based on the physics simulation;
- labeling the raw synthetic sensor data with ground truth parameters;
- and
- based on the labeled raw synthetic sensor data, outputting a labeled training dataset for training a machine learning model.
A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform operations comprising:
- defining, within a digital twin simulation environment, a virtual container having specified geometry and material properties;
- configuring one or more virtual sensors within the digital twin simulation environment;
- executing a physics simulation to model electromagnetic wave propagation through the virtual container and interaction with a virtual enclosed medium;
- generating raw synthetic sensor data based on the physics simulation;
- labeling the raw synthetic sensor data with ground truth parameters;
- and
- based on the labeled raw synthetic sensor data, outputting a labeled training dataset for training a machine learning model.
A system for generating synthetic training data for a container monitoring system, comprising:
- a physical sensor configured for external mounting to a physical container;
- and
- one or more processors configured to perform operations comprising:
- defining, within a digital twin simulation environment, a virtual container having specified geometry and material properties;
- configuring one or more virtual sensors within the digital twin simulation environment;
- executing a physics simulation to model electromagnetic wave propagation through the virtual container and interaction with a virtual enclosed medium;
- generating raw synthetic sensor data based on the physics simulation;
- labeling the raw synthetic sensor data with ground truth parameters;
- based on the labeled raw synthetic sensor data, outputting a labeled training dataset for training a machine learning model;
- collecting real-world sensor data from the physical sensor monitoring the physical container;
- identifying a domain gap between the raw synthetic sensor data and the real-world sensor data;
- and
- updating the digital twin simulation environment to reduce the domain gap.
Dependencies are inferred from the claim text.