Path Estimation Learning Using Reception-Level Sequences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cell production systems, the low accuracy of position estimation using fingerprint systems makes it difficult to associate process-in-operation information with worker positions, hindering the ability to track processes, order, and paths within the system.
Innovation Solution
A learning apparatus and path estimation system that collect and associate process-in-operation information with reception level information to learn a generation model, enabling accurate estimation of worker movements using a Conditional Variational AutoEncoder (CVAE) for improved position estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fingerprint system is used for position estimation, then the system complexity is reduced and ease of operation is improved, but the position estimation accuracy deteriorates making it difficult to associate process-in-operation information with worker positions
Solution Approach 1:
The patent changes the parameters used for position estimation from simple radio wave intensity values to sequences of reception levels over time. By incorporating temporal information and using multiple reception level measurements, the system achieves higher estimation accuracy while maintaining the simplicity of the fingerprint system approach. The generation model learns to estimate position based on sequences of reception levels rather than single-point measurements.
2Productivity
If multiple workers are present in a narrow area, then productivity is improved through efficient space utilization, but the difficulty of detecting and measuring individual worker positions increases due to mixed process-in-operation information
Solution Approach 1:
The patent segments the mixed process-in-operation information by using reception level sequences as unique identifiers for each worker. The generation model learns to associate specific reception level patterns with individual workers, effectively segmenting the mixed information from multiple workers in narrow areas. This allows individual worker tracking even when multiple workers are present simultaneously in the same physical space.
Solution Approach 2:
The system uses feedback from the generation model's learning process to improve position detection. The model is trained on historical data containing reception level sequences and actual worker positions, continuously refining its ability to distinguish between multiple workers. This learned feedback mechanism enables accurate identification of individual workers' positions even when their information is mixed in the sensor data.
Data Source
AI summary
This learning device is provided with: collection circuits (105, 106) for collecting step operation information indicating the operational state of each step, and reception level information that includes a reception level of a signal received by a first wireless station (104) from each of second wireless stations (103-1 to 103-3) installed for each step; and a learning circuit (108) for using teaching data formed of the step operation information and the reception level information linked to each other for each step to cause a generation model for estimating the operational state of a third wireless station (102) to learn.