Robot Work Factor Analysis Using Sensor Contribution Data

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems struggle to accurately identify various failure factors that cause inappropriate work by work machines due to their reliance on predefined determinations, limiting their ability to handle diverse environmental and situational factors.

Innovation Solution

A factor analysis device utilizing an input unit, learning unit, and analysis unit to construct an estimation model through machine learning, analyzing sensor data and determination results to determine the contribution degree of sensor data to work stoppages or quality deterioration, and notify the findings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If predefined failure factors are used for determination, then the system can clearly identify specific failure causes, but the system cannot adequately identify diverse failure factors that occur in various environments

Engineering Contradiction:
Improvefailure factor identification accuracyVSAvoidability to handle diverse failure factors
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system changes from using fixed predefined failure factors to dynamically generated failure factors based on machine learning model predictions. The failure factors are derived from input data that the model determines is important for predicting work stoppage or quality deterioration, allowing the system to adapt to diverse environments while maintaining precise identification through the contribution degree analysis.

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If machine learning estimation model is constructed with multiple sensor data, then the system can identify diverse failure factors, but the system complexity increases

Engineering Contradiction:
Improveability to identify diverse failure factorsVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The machine learning estimation model acts as an intermediary that processes multiple sensor data inputs and translates them into interpretable failure factor identifications. The model serves as a mediator between complex sensor data and meaningful failure factor conclusions, managing system complexity while enabling diverse failure factor identification through its internal processing mechanisms.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system replaces traditional mechanical or rule-based analysis methods with a machine learning-based estimation model. This substitution allows the system to handle diverse failure factors automatically without requiring explicit programming for each scenario, reducing operational complexity while maintaining high adaptability through data-driven pattern recognition.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If contribution degree analysis is performed on sensor data, then the system can accurately identify failure factors, but the analysis time and computational resources increase

Engineering Contradiction:
Improvefailure factor identification accuracyVSAvoidanalysis time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The machine learning estimation model performs preliminary analysis of sensor data during the training and prediction phases, pre-identifying which input data elements are most important for predicting work stoppage or quality deterioration. This preliminary action enables the subsequent contribution degree analysis to focus only on relevant data, reducing the time and computational resources needed for accurate failure factor identification.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4597246A1Factor analysis device and factor analysis method
Publication Date: 2025.08.06 KAWASAKI JUKOGYO KK
  • EP4597246A1 patent drawingFigure 1
  • EP4597246A1 patent drawingFigure 2
  • EP4597246A1 patent drawingFigure 3~4

AI summary

A factor analysis device (50) includes an input unit (51), a learning unit (52a), an analysis unit (52b), and a notification unit (52c). The input unit (51) receives sensor data obtained by measuring a work performed by a robot (11) with a sensor, and a determination result of a work stoppage or a work quality of a work corresponding to the sensor data. The learning unit (52a) constructs a estimation model by machine learning based on input data to the input unit (51a), with the sensor data as input data and an estimation result of the work stoppage or the work quality as output data. The analysis unit (52b) analyzes the input data and the output data of the estimation model to create the contribution data indicating the contribution degree that is a degree to which the sensor data contributes to an estimation of the work stoppage or the work quality. The notification unit (52c) notifies the contribution data created by the analysis unit (52b).