Storage Device Identification Model Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage devices struggle to accurately identify objects across different environments without RFID tags, as not all goods are tagged, and RFID scanning devices are not widely popular, necessitating a solution for reliable object identification in various settings.
Innovation Solution
A storage device comprising a storage space, sensor, processor, and transceiver that generates and uploads an identification model based on sensing data to a server, allowing other devices to identify objects using this model, facilitating accurate and efficient identification even across different configurations and environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If RFID tags are used for object identification, then identification accuracy is improved, but device complexity and cost increase due to requiring RFID scanning devices and tags on all goods
Solution Approach 1:
The patent extracts the identification model from the complex RFID system and stores it separately in a database. The sensor only needs to capture basic sensing data, while the pre-trained identification model performs the complex pattern recognition, thereby simplifying the device structure while maintaining high identification accuracy.
Solution Approach 2:
The identification model is pre-trained using sensing data collected in advance under various environmental conditions. This preliminary training allows the model to be stored in the database and reused for identification without requiring real-time complex processing or RFID infrastructure, reducing both device complexity and cost.
2Reliability
If RFID scanning devices are deployed widely, then object tracking capability is improved, but implementation cost and complexity increase
Solution Approach 1:
Instead of deploying expensive RFID scanning devices, the patent creates a copy of the identification capability through the identification model stored in the database. Any device with basic sensors can use this pre-trained model to achieve reliable object tracking without requiring specialized RFID hardware, thereby reducing implementation complexity and cost.
3Measurement precision
If sensors capture detailed sensing data for identification, then identification accuracy is improved, but computation load increases
Solution Approach 1:
The computationally intensive task of pattern recognition is performed in advance during the training phase, where the identification model learns from detailed sensing data. During actual operation, only lightweight inference is required using the pre-trained model, significantly reducing real-time computation load and energy consumption while maintaining high identification accuracy.
4Measurement precision
If identification models are generated for each storage device, then identification accuracy in specific environments is improved, but system complexity and data management burden increase
Solution Approach 1:
The patent creates a universal identification model that can be applied across multiple storage devices and environments. The model is trained on diverse sensing data to become environment-agnostic, allowing a single model stored in the database to serve multiple devices, thereby reducing system complexity and data management burden while maintaining high identification accuracy across different settings.
Data Source
AI summary
Provided is a storage method including the following. An object is stored in a storage space. The object is sensed to generate sensing data. An identification model is generated according to the sensing data. In addition, the identification model is uploaded to a server.