A multi-modal data injection defense method for heterogeneous embodied intelligent devices
By filtering data from embodied intelligent devices, implementing real-time priority scheduling, and quantifying cross-modal features, the problem of attack identification and self-healing of embodied intelligent devices in extreme environments has been solved, achieving stable operation and secure response.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TAIZHOU RES INST ZHEJIANG UNIV OF TECH
- Filing Date
- 2026-05-28
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies lack quantifiable execution standards, edge computing power scheduling adaptation, anti-interference capabilities, and complex attack tracing in embodied intelligent devices, resulting in high risks of false alarms, missed alarms, system failures, and secondary accidents.
By processing data through filtering and parameter compensation, implementing a real-time priority scheduling model, cross-modal feature quantization fusion, and adaptive recovery strategies, we can identify and respond to data injection attacks.
It enables embodied intelligent devices to operate stably in extreme environments, reduces false alarm rates, identifies complex attacks, prevents frequent system anomalies, provides self-healing capabilities, and avoids secondary accidents.
Smart Images
Figure CN122419953A_ABST