Generic Work Recognition Model Generation for Manufacturing Adaptability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing work recognition systems lack versatility in recognizing new processes or products, requiring changes to recognition logic and learning models, which is time-consuming.
Innovation Solution
A work recognition system that includes a storage unit for work component recognition models, a unit work recognition model generation unit, and a unit work recognition unit, enabling generic unit work recognition independent of products.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If special-purpose recognition logic and learning models are introduced for work elements, then work recognition accuracy is improved, but system adaptability to new processes and products deteriorates
Solution Approach 1:
The patent segments work recognition into hierarchical levels: work element recognition (basic actions like grasping, moving), work recognition (specific task actions), and process recognition (overall workflow). This segmentation allows generic work element models to be combined flexibly for different specific tasks, improving both accuracy through specialized element models and adaptability through modular composition.
Solution Approach 2:
The patent creates universal work element recognition models that can be applied across multiple work contexts. By defining generic work elements (grasping, moving, positioning) that are independent of specific products or processes, the system achieves versatility while maintaining recognition accuracy through specialized element models.
2Measurement precision
If special-purpose recognition logic and learning models are introduced for work elements, then work recognition accuracy is improved, but system complexity and preparation time deteriorate
Solution Approach 1:
The patent divides the recognition system into independent work element models that can be developed, validated, and maintained separately. This segmentation reduces system complexity by allowing modular development of recognition logic while maintaining high accuracy through specialized element models.
Solution Approach 2:
The patent performs preliminary action by pre-defining generic work element recognition models that can be reused across different tasks. This preliminary preparation of universal element models reduces the complexity of creating new recognition systems for different processes, as only task-specific combinations need to be defined rather than complete recognition logic.
3Measurement precision
If special-purpose recognition logic and learning models are introduced for work elements, then work recognition accuracy is improved, but time required for preparing new recognition systems deteriorates
Solution Approach 1:
The patent creates universal work element recognition models that can be applied across multiple work contexts without modification. This universality eliminates the need to create new recognition logic for each product or process, significantly reducing preparation time while maintaining accuracy through the specialized element models.
Solution Approach 2:
The patent performs preliminary action by pre-defining generic work element recognition models that can be reused across different tasks. This preliminary preparation of universal element models allows rapid deployment for new processes and products, reducing preparation time while maintaining recognition accuracy.
Data Source
AI summary
A work recognition system including: a storage unit configured to store a plurality of work component recognition models that take as input information of movement, touch, hearing, position, camera, and equipment, and tools at the time of a work to an object of a worker, divide the work into work components, and make a determination; a unit work recognition model generation unit configured to generate a general-purpose unit work recognition model by combining components independent of products among the work component recognition models based on work contents; and a unit work recognition unit configured to input, to the unit work recognition model, information of movement, touch, hearing, position, camera, and equipment, and tools at the time of a work to an object of a worker, and recognizing a unit work is provided.