NLU-based systems and method for the facilitated control of industrial assets
AssigneeSiemens Aktiengesellschaft
InventorsDaniel Krüger, Florian Kubo, Martina Schubert
Life of this patent
Estimating the term…
Abstract
A computer-executable Natural Language Understanding configuration generator is configured to generate a Natural Language Understanding configuration. The Natural Language Understanding configuration for a Natural Language Understanding component is based on engineering data related to an asset of an automation system.
Claims11
A voice control system of a user interface of an asset of an automation system, the voice control system comprising:
- a natural language understanding (NLU) component comprising an NLU configuration;
- and
- an NLU configuration generator that automatically generates the NLU configuration based on engineering data that is related to the asset of the automation system and that is taken from an automation or engineering project, wherein the engineering data comprises data related to a structure and/or functional structure and/or user interfaces of the asset and wherein the NLU configuration generator is configured to generate annotated examples and descriptions of the data related to the structure and/or functional structure and/or user interfaces of the asset when generating the NLU configuration from the engineering data, wherein the annotated examples comprise sentences, wherein intents, entities and entity roles are identified in the sentences, and the descriptions comprise value ranges and/or RegEx.
A computer-implemented method for configuring an NLU component of a voice control system of an asset of an automation system, the computer implemented method comprising:
- automatically generating an NLU configuration of the NLU component based on engineering data relating to the asset and taken from an automation or engineering project, with the engineering data comprising data related to a structure and/or functional structure and/or user interfaces of the asset;
- configuring the generated NLU configuration to generate annotated examples and descriptions of the data related to the structure and/or functional structure and/or user interfaces of the asset when generating the NLU configuration from the engineering data, wherein the annotated examples comprise sentences;
- and
- identifying Intents, entities and entity roles in the sentences and the descriptions comprise value ranges and/or RegEx.
A method for voice-assisted control of an asset of an automation system, the method comprising:
- automatically generating an NLU configuration of an NLU component of a voice control program based on engineering data relating to the asset and taken from an automation or engineering project, with the engineering data comprising data related to a structure and/or functional structure and/or user interfaces of the asset;
- configuring the generated NLU configuration to generate annotated examples and descriptions of the data related to the structure and/or functional structure and/or user interfaces of the asset when generating the NLU configuration from the engineering data, wherein the annotated examples comprise sentences;
- identifying intents, entities and entity roles in the sentences, and the descriptions comprise value ranges and/or RegEx;
- receiving an audio input generated by a user of the asset of the automation system with the voice control program;
- generating text from the audio input with an automatic speech recognition component of the voice control program;
- generating a corresponding semantic representation from the text with the NLU component of the voice control program;
- generating a machine-readable query to the asset of the automation system from the semantic representation of the text with a dialog manager component of the voice control program;
- forwarding the machine-readable query as a functional query to the asset of the automation system by the dialog manager of the voice control program;
- and
- controlling the asset of the automation system according to the functional query.
Dependencies are inferred from the claim text.