Distributed anti-interruption cooperative navigation method based on generative graph adaptive filtering
By generating virtual observation and prediction covariance matrices using a masked spatiotemporal graph Transformer model and combining it with adaptive Kalman filtering, the problem of error accumulation caused by link interruption in distributed cooperative navigation systems is solved, achieving high-precision navigation estimation and robustness.
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THE 54TH RESEARCH INSTITUTE OF CHINA ELECTRONICS TECHNOLOGY GROUP CORPORATION
- Filing Date
- 2026-01-28
- Publication Date
- 2026-05-29
AI Technical Summary
In complex environments where satellite navigation signals are denied, millimeter-wave and terahertz signals in distributed cooperative navigation systems are easily blocked, leading to sudden interruptions in communication and measurement links. Traditional retransmission mechanisms fail, and missing observations result in error accumulation. Existing deep learning methods lack group perception and uncertainty quantification, leading to error cascading.
A masked spatiotemporal graph Transformer model is used to generate virtual observation and prediction covariance matrices. Combined with adaptive Kalman filtering, adaptive measurement pairs are constructed, and uncertainty perception is used to suppress error cascading, thereby achieving high-precision navigation.
In the event of a link interruption, high-precision virtual observations are generated to suppress error cascading, maintain the consistency of navigation estimation, dynamically adjust filter weights, and improve the robustness and accuracy of the navigation system.
Smart Images

Figure CN122108116A_ABST