Motion Analysis Apparatus Using Waveform Similarity for Automatic Labeling

Solution Overview

Problem

Existing motion analysis methods require significant time and effort to assign motion labels to time-series data from sensors, and there is a need for automatic labeling to compile worker motion history and analyze movement effectively.

Innovation Solution

A motion analysis apparatus that detects corresponding segments in time-series data from worker and object sensors using waveform similarity, associates this information, and generates learning data for constructing an estimator to automate motion labeling and analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a motion dictionary is used to specify worker motion by comparing sensor output with predefined motion patterns, then motion analysis can be performed, but significant time and effort are required to assign motion labels to time-series data for constructing the dictionary

Engineering Contradiction:

Improvemotion specification accuracyVSAvoidtime for labeling time-series data

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system automatically generates learning data by detecting corresponding segments between worker sensor data and object sensor data, eliminating the need for manual motion labeling. The worker's motion is inferred from the object's motion pattern, allowing the system to self-generate training data without human intervention.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The object's motion data serves as an intermediary to indirectly obtain the worker's motion information. By detecting corresponding segments between the worker sensor data and object sensor data, the system uses the object's motion as a mediator to automatically label the worker's motion without direct observation or manual annotation.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If manual motion labeling is performed to construct a motion dictionary, then reliable motion data can be obtained, but the process requires significant effort and cannot be automated

Engineering Contradiction:

Improvemotion data reliabilityVSAvoidautomation of motion labeling

Core Design Contradiction:

ReliabilityVSExtent of automation

Solution Approach 1:

The system automatically generates learning data by detecting corresponding segments between worker sensor data and object sensor data, eliminating the need for manual motion labeling. The worker's motion is inferred from the object's motion pattern, allowing the system to self-generate training data without human intervention.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The object's motion data serves as an intermediary to indirectly obtain the worker's motion information. By detecting corresponding segments between the worker sensor data and object sensor data, the system uses the object's motion as a mediator to automatically label the worker's motion without direct observation or manual annotation.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Quantity of substance

If sensors are attached to workers to measure movement in the field, then motion history can be compiled, but the complexity of data processing and analysis increases

Engineering Contradiction:

Improvemotion data quantityVSAvoiddata processing complexity

Core Design Contradiction:

Quantity of substanceVSDevice complexity

Solution Approach 1:

The system extracts only the relevant corresponding segments from the large volume of sensor data by comparing waveform patterns between worker and object sensors. This extraction approach filters out unnecessary data and focuses only on the segments that contain meaningful motion information, simplifying subsequent processing.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The object's motion data serves as an intermediary to indirectly obtain the worker's motion information. By detecting corresponding segments between the worker sensor data and object sensor data, the system uses the object's motion as a mediator to automatically label the worker's motion without direct observation or manual annotation.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11030564B2Motion analysis apparatus, motion analysis method, and computer program product

Publication Date: 2021.06.08 KK TOSHIBA

AI summary

According to an embodiment, a motion analysis apparatus includes a memory and processing circuitry. The processing circuitry detects a corresponding segment using first time-series data based on output of a first sensor configured to measure movement of a first object and second time-series data based on output of a second sensor configured to measure movement of a second object, the corresponding segment being a segment in which the first time-series data and the second time-series data are similar in their waveform patterns or co-occur. The processing circuitry associates information specifying the detected corresponding segment with at least one of information specifying the first object or information specifying the second object.