Nonverbal Information Generation for Communication Robots

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing techniques for generating nonverbal information for communication robots and agents require high costs due to the need for manual creation and registration of utterance and nonverbal action pairs, making automation of voice and text association with nonverbal information challenging.

Innovation Solution

A nonverbal information generation apparatus that uses a learned model to generate time-information-stamped nonverbal information based on time-information-stamped text feature quantities, allowing for automated association of voice and text information with nonverbal behavior, reducing the need for manual registration and lowering production costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If manual creation and registration of utterance and nonverbal action pairs is performed, then the quality and accuracy of nonverbal information can be ensured, but the cost and time consumption increase significantly

Engineering Contradiction:

Improvequality of nonverbal informationVSAvoidtime consumption for data creation

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent uses machine learning models to automatically generate nonverbal information by learning from existing data patterns, effectively copying successful nonverbal behaviors from training examples rather than manually creating each utterance-action pair. This automation maintains quality through learned patterns while dramatically reducing manual time investment.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system performs self-service by automatically generating nonverbal information through the learned model without requiring manual annotation for each new utterance. The model serves itself by processing new text inputs and generating corresponding nonverbal actions autonomously, eliminating the need for continuous manual data creation.

Inventive Principle:

Principle #25Self-service

2Adaptability or versatility

If manual creation and registration of utterance and nonverbal action pairs is performed, then comprehensive nonverbal coverage can be achieved, but the production cost increases

Engineering Contradiction:

Improvecoverage of nonverbal informationVSAvoidproduction cost

Core Design Contradiction:

Adaptability or versatilityVSEase of manufacture

Solution Approach 1:

The learned model serves multiple functions: it generates nonverbal information for various types of utterances, adapts to different communication contexts, and can be applied across different applications. This universal approach achieves comprehensive coverage without the prohibitive costs of manually creating specialized datasets for each scenario.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system achieves versatility by changing parameters such as adjusting the model's generation behavior based on different input types, confidence thresholds, and contextual factors. This allows comprehensive coverage of various nonverbal scenarios through parameter adjustment rather than manual creation of each scenario.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If automated generation of nonverbal information is implemented, then cost and time efficiency improve, but the naturalness and human-like quality may deteriorate

Engineering Contradiction:

Improveefficiency of nonverbal information generationVSAvoidnaturalness of nonverbal behavior

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system incorporates feedback mechanisms where the generated nonverbal information can be evaluated and refined. The model learns from feedback signals during training and can adjust its generation to improve naturalness, ensuring that automated production maintains high quality and human-like characteristics.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The generated nonverbal information is dynamic and adaptable, changing based on the input context, timing, and learned patterns from diverse training data. This dynamic generation produces natural, human-like variations rather than static, repetitive responses, maintaining reliability while achieving high productivity.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS11404063B2Nonverbal information generation apparatus, nonverbal information generation model learning apparatus, methods, and programs

Publication Date: 2022.08.02 NIPPON TELEGRAPH & TELEPHONE CORP

AI summary

A nonverbal information generation apparatus includes a nonverbal information generation unit that generates time-information-stamped nonverbal information that corresponds to time-information-stamped text feature quantities and an expression unit that expresses the nonverbal information on the basis of the time-information-stamped text feature quantities and a learned nonverbal information generation model. The time-information-stamped text feature quantities are configured to include feature quantities that have been extracted from text and time information representing times assigned to predetermined units of the text. The nonverbal information is information for controlling the expression unit so as to express behavior corresponding to the text. The nonverbal information generation unit controls the expression unit so that the time-information-stamped nonverbal information is expressed from the expression unit in accordance with time information assigned to the nonverbal information, and controls the expression unit so that the text or voice corresponding to the text that corresponds to the time information is output.