基于区块链与联邦学习的具身设备数据管理系统和方法
By combining blockchain with federated learning, the security and ownership issues in embodied device data management are resolved, enabling secure data flow and privacy protection, incentivizing the contribution of high-quality data, and optimizing the performance of the global model and the intelligent agent ecosystem.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU FANJIA TECH CO LTD
- Filing Date
- 2026-04-16
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, the management of embodied device data suffers from low security in centralized data aggregation mode and ambiguous ownership in decentralized isolated storage mode, leading to data isolation and privacy concerns, and inhibiting the sharing and joint analysis of high-quality data.
The data management system adopts blockchain and federated learning. It collects and extracts multimodal data on the device, generates data asset packages, adds digital fingerprints and timestamps, writes them to the blockchain for notarization, performs federated learning and privacy computing locally to optimize the global model, and incentivizes data contributors by paying digital tokens through smart contracts.
It has achieved secure data flow and privacy protection, clarified the rights and interests of data creators, incentivized the contribution of high-quality data, and optimized the performance of the global model and the level of the intelligent agent ecosystem.
Smart Images
Figure CN122046414B_ABST
Abstract
Citation Information
Patent Citations
CN121525081A